diff --git a/src/univariate/discrete/bernoulli.jl b/src/univariate/discrete/bernoulli.jl
index f0d812e5b..8d613fe45 100644
--- a/src/univariate/discrete/bernoulli.jl
+++ b/src/univariate/discrete/bernoulli.jl
@@ -1,13 +1,11 @@
 immutable Bernoulli <: DiscreteUnivariateDistribution
     p::Float64
 
-    function Bernoulli(p::Float64)
-        0.0 <= p <= 1.0 ||
+    function Bernoulli(p::Real)
+        zero(p) <= p <= one(p) ||
             throw(ArgumentError("Bernoulli: p must be in [0, 1]."))
-        new(p)
+        @compat new(Float64(p))
     end
-
-    Bernoulli(p::Real) = @compat Bernoulli(Float64(p))
     Bernoulli() = new(0.5)
 end
 
diff --git a/src/univariate/discrete/categorical.jl b/src/univariate/discrete/categorical.jl
index 6731aac28..2512812c5 100644
--- a/src/univariate/discrete/categorical.jl
+++ b/src/univariate/discrete/categorical.jl
@@ -5,12 +5,14 @@ immutable Categorical <: DiscreteUnivariateDistribution
     Categorical(p::Vector{Float64}, ::NoArgCheck) = new(length(p), p)
 
     function Categorical(p::Vector{Float64})
-        isprobvec(p) || error("p is not a valid probability vector.")
+        isprobvec(p) ||
+            throw(ArgumentError("Categorical: p is not a valid probability vector."))
         new(length(p), p)
     end
 
     function Categorical(k::Integer)
-        k >= 1 || error("k must be a positive integer.")
+        k >= 1 ||
+            throw(ArgumentError("k must be a positive integer."))
         new(k, fill(1.0/k, k))
     end
 end
diff --git a/src/univariate/discrete/discreteuniform.jl b/src/univariate/discrete/discreteuniform.jl
index 110c7e0c4..90f5f2ce3 100644
--- a/src/univariate/discrete/discreteuniform.jl
+++ b/src/univariate/discrete/discreteuniform.jl
@@ -3,13 +3,15 @@ immutable DiscreteUniform <: DiscreteUnivariateDistribution
     b::Int
     pv::Float64
 
-    function DiscreteUniform(a::Int, b::Int)
-        a <= b || error("a and b must satisfy a <= b")
+    function DiscreteUniform(a::Real, b::Real)
+        a <= b ||
+            throw(ArgumentError("DiscreteUniform: a and b must satisfy a <= b."))
+        @compat a_ = Int(a)
+        @compat b_ = Int(b)
         new(a, b, 1.0 / (b - a + 1))
     end
 
-    @compat DiscreteUniform(a::Real, b::Real) = DiscreteUniform(round(Int, a), round(Int, b))
-    DiscreteUniform(b::Real) = DiscreteUniform(0, round(Int, b))
+    DiscreteUniform(b::Real) = DiscreteUniform(0, b)
     DiscreteUniform() = new(0, 1, 0.5)
 end
 
diff --git a/src/univariate/discrete/geometric.jl b/src/univariate/discrete/geometric.jl
index d77901ac7..f58483279 100644
--- a/src/univariate/discrete/geometric.jl
+++ b/src/univariate/discrete/geometric.jl
@@ -2,11 +2,12 @@ immutable Geometric <: DiscreteUnivariateDistribution
     p::Float64
 
     function Geometric(p::Real)
-        zero(p) < p < one(p) || error("prob must be in (0, 1)")
+        zero(p) < p < one(p) ||
+            throw(ArgumentError("Geometric: p must be in (0, 1)."))
     	@compat new(Float64(p))
     end
 
-    Geometric() = Geometric(0.5) # Flips of a fair coin
+    Geometric() = new(0.5) # Flips of a fair coin
 end
 
 @distr_support Geometric 0 Inf
diff --git a/src/univariate/discrete/hypergeometric.jl b/src/univariate/discrete/hypergeometric.jl
index 8845eb54f..ea9e2339c 100644
--- a/src/univariate/discrete/hypergeometric.jl
+++ b/src/univariate/discrete/hypergeometric.jl
@@ -11,14 +11,13 @@ immutable Hypergeometric <: DiscreteUnivariateDistribution
     nf::Int     # number of failures in population
     n::Int      # sample size
 
-    function Hypergeometric(ns::Int, nf::Int, n::Int)
-        ns >= 0 || error("ns must be non-negative.")
-        nf >= 0 || error("nf must be non-negative.")
-        0 < n < ns + nf || error("n must have 0 < n < ns + nf")
-        new(ns, nf, n)
+    function Hypergeometric(ns::Real, nf::Real, n::Real)
+        (ns >= 0 && nf >= 0) ||
+            throw(ArgumentError("Hypergeometric: ns must be non-negative."))
+        0 < n < ns + nf ||
+            throw(ArgumentError("Hypergeometric: n must have 0 < n < ns + nf."))
+        @compat new(Int(ns), Int(nf), Int(n))
     end
-
-    @compat Hypergeometric(ns::Real, nf::Real, n::Real) = Hypergeometric(round(Int, ns), round(Int, nf), round(Int, n))
 end
 
 @distr_support Hypergeometric max(d.n - d.nf, 0) min(d.ns, d.n)
diff --git a/src/univariate/discrete/negativebinomial.jl b/src/univariate/discrete/negativebinomial.jl
index 5cbde211a..5ca9674f0 100644
--- a/src/univariate/discrete/negativebinomial.jl
+++ b/src/univariate/discrete/negativebinomial.jl
@@ -8,15 +8,16 @@ immutable NegativeBinomial <: DiscreteUnivariateDistribution
     r::Int
     p::Float64
 
-    function NegativeBinomial(r::Int, p::Float64)
-        r > 0 || error("r must be positive.")
-        0.0 < p <= 1.0 || error("prob must be in (0, 1].")
-        new(r, p)
+    function NegativeBinomial(r::Real, p::Real)
+        r > zero(r) ||
+            throw(ArgumentError("NegativeBinomial: r must be positive."))
+        zero(p) < p <= one(p) ||
+            throw(ArgumentError("NegativeBinomial: prob must be in (0, 1]."))
+        @compat new(Int(r), Float64(p))
     end
 
-    @compat NegativeBinomial(r::Real, p::Real) = NegativeBinomial(round(Int, r), Float64(p))
     NegativeBinomial(r::Real) = NegativeBinomial(r, 0.5)
-    NegativeBinomial() = new(1.0, 0.5)
+    NegativeBinomial() = new(1, 0.5)
 end
 
 @distr_support NegativeBinomial 0 Inf
diff --git a/src/univariate/discrete/noncentralhypergeometric.jl b/src/univariate/discrete/noncentralhypergeometric.jl
index c04cdf863..3766897c8 100644
--- a/src/univariate/discrete/noncentralhypergeometric.jl
+++ b/src/univariate/discrete/noncentralhypergeometric.jl
@@ -1,4 +1,5 @@
 # Noncentral hypergeometric distribution
+# TODO: this distribution needs clean-up and testing
 
 abstract NoncentralHypergeometric <: DiscreteUnivariateDistribution
 
@@ -104,6 +105,3 @@ function pdf(d::WalleniusNoncentralHypergeometric, k::Int)
 end
 
 logpdf(d::WalleniusNoncentralHypergeometric, k::Int) = log(pdf(d, k))
-
-
-
diff --git a/src/univariate/discrete/poisson.jl b/src/univariate/discrete/poisson.jl
index c0edc5867..1726dced1 100644
--- a/src/univariate/discrete/poisson.jl
+++ b/src/univariate/discrete/poisson.jl
@@ -1,12 +1,11 @@
 immutable Poisson <: DiscreteUnivariateDistribution
     λ::Float64
 
-    function Poisson(λ::Float64)
-        λ > 0.0 || error("λ must be positive.")
-        new(λ)
+    function Poisson(λ::Real)
+        λ > zero(λ) ||
+            throw(ArgumentError("Poisson: λ must be positive."))
+        @compat new(Float64(λ))
     end
-
-    @compat Poisson(λ::Real) = Poisson(Float64(λ))
     Poisson() = new(1.0)
 end
 
diff --git a/src/univariate/discrete/poissonbinomial.jl b/src/univariate/discrete/poissonbinomial.jl
index 1c8e7d73d..afa19394f 100644
--- a/src/univariate/discrete/poissonbinomial.jl
+++ b/src/univariate/discrete/poissonbinomial.jl
@@ -1,3 +1,4 @@
+# TODO: this distribution may need clean-up
 
 # Computes the pdf of a poisson-binomial random variable using
 # fast fourier transform
diff --git a/src/univariate/discrete/skellam.jl b/src/univariate/discrete/skellam.jl
index d5fdd3a5c..4522faac1 100644
--- a/src/univariate/discrete/skellam.jl
+++ b/src/univariate/discrete/skellam.jl
@@ -2,21 +2,18 @@ immutable Skellam <: DiscreteUnivariateDistribution
     μ1::Float64
     μ2::Float64
 
-    function Skellam(μ1::Float64, μ2::Float64)
-        μ1 > 0.0 && μ2 > 0.0 || error("μ1 and μ2 must be positive.")
-        new(μ1, μ2)
+    function Skellam(μ1::Real, μ2::Real)
+        μ1 > zero(μ1) && μ2 > zero(μ2) ||
+            throw(ArgumentError("Skellam: μ1 and μ2 must be positive."))
+        @compat new(Float64(μ1), Float64(μ2))
     end
 
-    @compat Skellam(μ1::Real, μ2::Real) = Skellam(Float64(μ1), Float64(μ2))
-
     Skellam(μ::Real) = Skellam(μ, μ)
-
     Skellam() = new(1.0, 1.0)
 end
 
 @distr_support Skellam -Inf Inf
 
-
 ### Parameters
 
 params(d::Skellam) = (d.μ1, d.μ2)
@@ -55,4 +52,3 @@ end
 ### Sampling
 
 rand(d::Skellam) = rand(Poisson(d.μ1)) - rand(Poisson(d.μ2))
-