diff --git a/R/associators.R b/R/associators.R
index 8f67eef..76a86cd 100644
--- a/R/associators.R
+++ b/R/associators.R
@@ -19,9 +19,9 @@ function(name, class = NULL)
 }
 
 RWeka_build_associator <-
-function(x, control, name)
+function(x, control, name, weights = NULL)
 {
-    instances <- read_data_into_Weka(x)
+    instances <- read_data_into_Weka(x, weights)
 
     ## Build the associator.
     associator <- .jnew(name)
diff --git a/R/classifiers.R b/R/classifiers.R
index b689825..7b9a317 100644
--- a/R/classifiers.R
+++ b/R/classifiers.R
@@ -34,7 +34,7 @@ function(name, class = NULL, handlers = list())
     if(is.na(match("data", names(handlers))))
         handlers$data <- .default_data_handler_for_classifiers
         
-    out <- function(formula, data, subset, na.action,
+    out <- function(formula, data, subset, na.action, weights = NULL,
                     control = Weka_control(), options = NULL)
     {
         ## The "usual" way of creating a model frame from the call.
@@ -45,7 +45,7 @@ function(name, class = NULL, handlers = list())
         mf <- eval(mf, parent.frame())
 	
         .structure(c(RWeka_build_classifier(mf, control, name, handlers,
-                                            options),
+                                            options, weights),
                      list(call = mc, handlers = handlers,
                           levels = levels(mf[[1L]]),
                           terms = attr(mf, "terms"))),
@@ -55,7 +55,7 @@ function(name, class = NULL, handlers = list())
 }
 
 RWeka_build_classifier <-
-function(mf, control, name, handlers, options)
+function(mf, control, name, handlers, options, weights = NULL)
 {
     out <- list()
 
@@ -67,7 +67,7 @@ function(mf, control, name, handlers, options)
         out$model <- mf
 
     mf <- .compose_and_funcall(handlers$data, mf)
-    instances <- read_model_frame_into_Weka(mf)
+    instances <- read_model_frame_into_Weka(mf, weights)
 
     ## Build the classifier.
     classifier <- .jnew(name)
@@ -239,10 +239,10 @@ function(formula, ...)
 {
     if(!is.null(mf <- formula$model)) return(mf)
     dots <- list(...)
-    nargs <- dots[match(c("data", "na.action", "subset"),
+    nargs <- dots[match(c("data", "na.action", "subset", "weights"),
                         names(dots), 0L)]
     mf <- formula$call
-    mf <- mf[c(1L, match(c("formula", "data", "subset", "na.action"),
+    mf <- mf[c(1L, match(c("formula", "data", "subset", "na.action", "weights"),
                          names(mf), 0L))]
     mf$drop.unused.levels <- TRUE
     mf[[1L]] <- as.name("model.frame")
diff --git a/R/clusterers.R b/R/clusterers.R
index b021979..63c1326 100644
--- a/R/clusterers.R
+++ b/R/clusterers.R
@@ -19,9 +19,9 @@ function(name, class = NULL)
 }
 
 RWeka_build_clusterer <-
-function(x, control, name)
+function(x, control, name, weights = NULL)
 {
-    instances <- read_data_into_Weka(x)
+    instances <- read_data_into_Weka(x, weights)
 
     ## Build the clusterer.
     clusterer <- .jnew(name)
diff --git a/R/readers.R b/R/readers.R
index 45b8754..b06bae6 100644
--- a/R/readers.R
+++ b/R/readers.R
@@ -27,10 +27,10 @@ function(file)
 
 
 read_model_frame_into_Weka <-
-function(mf)
+function(mf, weights = NULL)
 {
     ## Model frame has the class variable in first position.
-    read_data_into_Weka(mf, 1L)
+    read_data_into_Weka(mf, weights, 1L)
 }
 
 write.arff <-
@@ -147,7 +147,7 @@ function(x)
 }
 
 read_data_into_Weka <-
-function(x, classIndex = 0L)
+function(x, weights = NULL, classIndex = 0L)
 {
     ## FastVector was deprecated in Weka >= 3-7-1. Now we have to use
     ## the List interface (see the cast of ArrayList in the Attribute
@@ -228,8 +228,12 @@ function(x, classIndex = 0L)
     ## Populate.
     x <- unlist(x, use.names = FALSE)
     x[is.na(x)] <- NaN                  # Weka missing value.
+    if(is.null(weights)) {
+        weights <- rep(1.0, n)
+    }
+    weights[is.na(weights)] <- 1.0
     .jcall("RWekaInterfaces", "V", "addInstances",
-           instances, .jarray(x), as.integer(n))
+           instances, .jarray(round(x, digits=6)), .jarray(weights), as.integer(n))
 
     ## Note that using dim and dimnames attributes would result in a
     ## matrix, which seems a bad idea.
diff --git a/java/RWekaInterfaces.java b/java/RWekaInterfaces.java
index 7d9530b..82db092 100644
--- a/java/RWekaInterfaces.java
+++ b/java/RWekaInterfaces.java
@@ -102,19 +102,19 @@ public abstract class RWekaInterfaces {
     // Instance is now an Interface, Weka >= 3-7-1. The equivalent
     // class seems to be DenseInstance.
     public static void addInstances(Instances instances, double[] data, 
-				    int nrow)
+				    double[] weights, int nrow)
 	throws Exception
     {
 	int i, j, ncol = instances.numAttributes();
 
-	if (data.length / ncol != nrow) {
+	if ((double) data.length / (double) ncol != nrow) {
 	    throw new Exception("invalid number of rows 'nrow'");
 	}
 
 	for (i = 0; i < nrow; i++) {
 	    Instance instance = new DenseInstance(ncol);
-	    for (j = 0; j < ncol; j++) 
-		instance.setValue(j, data[i+j*nrow]);
+	    for (j = 0; j < ncol; j++) { instance.setValue(j, data[i+j*nrow]); }
+        instance.setWeight(weights[i]);
 	    instances.add(instance);
 	}
     }
diff --git a/man/Weka_associators.Rd b/man/Weka_associators.Rd
index 549c6a3..39a9780 100644
--- a/man/Weka_associators.Rd
+++ b/man/Weka_associators.Rd
@@ -6,8 +6,8 @@
   R interfaces to Weka association rule learning algorithms.
 }
 \usage{
-Apriori(x, control = NULL)
-Tertius(x, control = NULL)
+Apriori(x, control = NULL, weights = NULL)
+Tertius(x, control = NULL, weights = NULL)
 }
 \arguments{
   \item{x}{an R object with the data to be associated.}
@@ -15,6 +15,8 @@ Tertius(x, control = NULL)
     character vector of control options, or \code{NULL} (default).
     Available options can be obtained on-line using the Weka Option
     Wizard \code{\link{WOW}}, or the Weka documentation.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \value{
   A list inheriting from class \code{Weka_associators} with components
diff --git a/man/Weka_classifier_functions.Rd b/man/Weka_classifier_functions.Rd
index e7f0fed..2e6f43e 100644
--- a/man/Weka_classifier_functions.Rd
+++ b/man/Weka_classifier_functions.Rd
@@ -9,11 +9,11 @@
 }
 \usage{
 LinearRegression(formula, data, subset, na.action,
-                 control = Weka_control(), options = NULL)
+                 control = Weka_control(), options = NULL, weights = NULL)
 Logistic(formula, data, subset, na.action,
-         control = Weka_control(), options = NULL)
+         control = Weka_control(), options = NULL, weights = NULL)
 SMO(formula, data, subset, na.action,
-    control = Weka_control(), options = NULL)
+    control = Weka_control(), options = NULL, weights = NULL)
 }
 \arguments{
   \item{formula}{a symbolic description of the model to be fit.}
@@ -29,6 +29,8 @@ SMO(formula, data, subset, na.action,
     the Weka documentation.}
   \item{options}{a named list of further options, or \code{NULL}
     (default).  See \bold{Details}.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \value{
   A list inheriting from classes \code{Weka_functions} and
diff --git a/man/Weka_classifier_lazy.Rd b/man/Weka_classifier_lazy.Rd
index fbd022c..be3ab01 100644
--- a/man/Weka_classifier_lazy.Rd
+++ b/man/Weka_classifier_lazy.Rd
@@ -8,9 +8,9 @@
 }
 \usage{
 IBk(formula, data, subset, na.action,
-    control = Weka_control(), options = NULL)
+    control = Weka_control(), options = NULL, weights = NULL)
 LBR(formula, data, subset, na.action,
-    control = Weka_control(), options = NULL)
+    control = Weka_control(), options = NULL, weights = NULL)
 }
 \arguments{
   \item{formula}{a symbolic description of the model to be fit.}
@@ -26,6 +26,8 @@ LBR(formula, data, subset, na.action,
     the Weka documentation.}
   \item{options}{a named list of further options, or \code{NULL}
     (default).  See \bold{Details}.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \value{
   A list inheriting from classes \code{Weka_lazy} and
diff --git a/man/Weka_classifier_meta.Rd b/man/Weka_classifier_meta.Rd
index 4f8f847..9c9dd85 100644
--- a/man/Weka_classifier_meta.Rd
+++ b/man/Weka_classifier_meta.Rd
@@ -12,17 +12,18 @@
 }
 \usage{
 AdaBoostM1(formula, data, subset, na.action,
-           control = Weka_control(), options = NULL)
+           control = Weka_control(), options = NULL, weights = NULL)
 Bagging(formula, data, subset, na.action,
-        control = Weka_control(), options = NULL)
+        control = Weka_control(), options = NULL, weights = NULL)
 LogitBoost(formula, data, subset, na.action,
-           control = Weka_control(), options = NULL)
+           control = Weka_control(), options = NULL, weights = NULL)
 MultiBoostAB(formula, data, subset, na.action,
-             control = Weka_control(), options = NULL)
+             control = Weka_control(), options = NULL, weights = NULL)
 Stacking(formula, data, subset, na.action,
-         control = Weka_control(), options = NULL)
+         control = Weka_control(), options = NULL, weights = NULL)
 CostSensitiveClassifier(formula, data, subset, na.action,
-                        control = Weka_control(), options = NULL)
+                        control = Weka_control(), options = NULL,
+                        weights = NULL)
 }
 \arguments{
   \item{formula}{a symbolic description of the model to be fit.}
@@ -42,6 +43,8 @@ CostSensitiveClassifier(formula, data, subset, na.action,
     function.}
   \item{options}{a named list of further options, or \code{NULL}
     (default).  See \bold{Details}.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \value{
   A list inheriting from classes \code{Weka_meta} and
diff --git a/man/Weka_classifier_rules.Rd b/man/Weka_classifier_rules.Rd
index 4c0d1d6..6326f00 100644
--- a/man/Weka_classifier_rules.Rd
+++ b/man/Weka_classifier_rules.Rd
@@ -10,13 +10,13 @@
 }
 \usage{
 JRip(formula, data, subset, na.action,
-     control = Weka_control(), options = NULL)
+     control = Weka_control(), options = NULL, weights = NULL)
 M5Rules(formula, data, subset, na.action,
-        control = Weka_control(), options = NULL)
+        control = Weka_control(), options = NULL, weights = NULL)
 OneR(formula, data, subset, na.action,
-     control = Weka_control(), options = NULL)
+     control = Weka_control(), options = NULL, weights = NULL)
 PART(formula, data, subset, na.action,
-     control = Weka_control(), options = NULL)
+     control = Weka_control(), options = NULL, weights = NULL)
 }
 \arguments{
   \item{formula}{a symbolic description of the model to be fit.}
@@ -32,6 +32,8 @@ PART(formula, data, subset, na.action,
     the Weka documentation.}
   \item{options}{a named list of further options, or \code{NULL}
     (default).  See \bold{Details}.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \value{
   A list inheriting from classes \code{Weka_rules} and
diff --git a/man/Weka_classifier_trees.Rd b/man/Weka_classifier_trees.Rd
index 57ffbb9..35df486 100644
--- a/man/Weka_classifier_trees.Rd
+++ b/man/Weka_classifier_trees.Rd
@@ -11,13 +11,13 @@
 }
 \usage{
 J48(formula, data, subset, na.action,
-    control = Weka_control(), options = NULL)
+    control = Weka_control(), options = NULL, weights = NULL)
 LMT(formula, data, subset, na.action,
-    control = Weka_control(), options = NULL)
+    control = Weka_control(), options = NULL, weights = NULL)
 M5P(formula, data, subset, na.action,
-    control = Weka_control(), options = NULL)
+    control = Weka_control(), options = NULL, weights = NULL)
 DecisionStump(formula, data, subset, na.action,
-              control = Weka_control(), options = NULL)
+              control = Weka_control(), options = NULL, weights = NULL)
 }
 \arguments{
   \item{formula}{a symbolic description of the model to be fit.}
@@ -33,6 +33,8 @@ DecisionStump(formula, data, subset, na.action,
     the Weka documentation.}
   \item{options}{a named list of further options, or \code{NULL}
     (default).  See \bold{Details}.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \value{
   A list inheriting from classes \code{Weka_tree} and
diff --git a/man/Weka_classifiers.Rd b/man/Weka_classifiers.Rd
index a45faad..5855c8d 100644
--- a/man/Weka_classifiers.Rd
+++ b/man/Weka_classifiers.Rd
@@ -12,12 +12,13 @@
 
   R interface functions to Weka classifiers are created by
   \code{\link{make_Weka_classifier}}, and have formals \code{formula},
-  \code{data}, \code{subset}, \code{na.action}, and \code{control}
-  (default: none), where the first four have the \dQuote{usual} meanings
-  for statistical modeling functions in R, and the last again specifies
-  the control options to be employed by the Weka learner.  By default,
-  the model formulae should only use the \samp{+} and \samp{-} operators
-  to indicate the variables to be included or not used, respectively.
+  \code{data}, \code{subset}, \code{na.action}, \code{control}
+  (default: none), and \code{weights} (default: \code{NULL}) where the first
+  four have the \dQuote{usual} meanings for statistical modeling functions in R,
+  the fifth again specifies the control options to be employed by the Weka
+  learner, and the last gives the weight attached to each row in data. By
+  default, the model formulae should only use the \samp{+} and \samp{-}
+  operators to indicate the variables to be included or not used, respectively.
 
   Objects created by these interfaces always inherit from class
   \code{Weka_classifier}, and have at least suitable \code{print},
diff --git a/man/Weka_clusterers.Rd b/man/Weka_clusterers.Rd
index e0341d6..9ec56e0 100644
--- a/man/Weka_clusterers.Rd
+++ b/man/Weka_clusterers.Rd
@@ -9,11 +9,11 @@
   R interfaces to Weka clustering algorithms.
 }
 \usage{
-Cobweb(x, control = NULL)
-FarthestFirst(x, control = NULL)
-SimpleKMeans(x, control = NULL)
-XMeans(x, control = NULL)
-DBScan(x, control = NULL)
+Cobweb(x, control = NULL, weights = NULL)
+FarthestFirst(x, control = NULL, weights = NULL)
+SimpleKMeans(x, control = NULL, weights = NULL)
+XMeans(x, control = NULL, weights = NULL)
+DBScan(x, control = NULL, weights = NULL)
 }
 \arguments{
   \item{x}{an R object with the data to be clustered.}
@@ -21,6 +21,8 @@ DBScan(x, control = NULL)
     character vector of control options, or \code{NULL} (default).
     Available options can be obtained on-line using the Weka Option
     Wizard \code{\link{WOW}}, or the Weka documentation.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \value{
   A list inheriting from class \code{Weka_clusterers} with components
diff --git a/man/Weka_interfaces.Rd b/man/Weka_interfaces.Rd
index c468ed3..b7a3c05 100644
--- a/man/Weka_interfaces.Rd
+++ b/man/Weka_interfaces.Rd
@@ -10,9 +10,9 @@
   available interfaces.
 }
 \usage{
-make_Weka_associator(name, class = NULL)
-make_Weka_classifier(name, class = NULL, handlers = list())
-make_Weka_clusterer(name, class = NULL)
+make_Weka_associator(name, class = NULL, weights = NULL)
+make_Weka_classifier(name, class = NULL, handlers = list(), weights = NULL)
+make_Weka_clusterer(name, class = NULL, weights = NULL)
 make_Weka_filter(name, class = NULL)
 list_Weka_interfaces()
 }
@@ -25,6 +25,8 @@ list_Weka_interfaces()
     representing associators, classifiers, and clusterers).}
   \item{handlers}{a named list of special handler functions, see
     \bold{Details}.}
+  \item{weights}{an optional list of weights, one for each row in data. If
+    \code{NULL}, each row is assigned the weight 1.}
 }
 \details{
   \code{make_Weka_associator} and \code{make_Weka_clusterer} create an R