diff --git a/src/algorithm.jl b/src/algorithm.jl
index 825adb2..41ae12a 100644
--- a/src/algorithm.jl
+++ b/src/algorithm.jl
@@ -112,11 +112,35 @@ algorithm `algo` with verbose level `verbose`.
 """
 function sample_scenarios end
 
+function sample_scenarios(sp::AbstractStochasticProgram, n::Int,
+    depth_max=1000, to::TimerOutput=TimerOutput(), verbose::Int=0)
+    scenarios = []
+    for i in 1:n
+        push!(scenarios,sample_scenario(sp, depth_max))
+    end
+    return scenarios
+end
+
+function sample_scenario(sp::AbstractStochasticProgram, depth_max=1000,
+    to::TimerOutput=TimerOutput(), verbose::Int=0)
+    node = get(sp,MasterNode())
+    s = Vector{:<AbstractTransition}
+    it = 0
+    while !is_leaf(sp,node) && it < depth_max
+        tr = get(sp,RandomTransition,node)
+        node = tr.Target
+        push!(s, tr)
+        it += 1
+    end
+    #TODO update timer output
+    return s
+end
+
 """
     compute_bounds(algo::AbstractAlgorithm, paths::AbstractPaths, verbose)
 
-Return a tuple `(z_UB, σ)` where z_UB reprensets the upperbound computed by the
-algorithm `algo` by using paths `paths` generated during the forward pass and σ
+Return a tuple `(z_UB, σ)` where z_UB represents the upperbound computed by Monte Carlo with the
+algorithm `algo` by using paths `paths` and σ
 represents the standard deviation of this upper bound.
 """
 function compute_bounds end
diff --git a/src/attributes.jl b/src/attributes.jl
index fb96a35..a02f377 100644
--- a/src/attributes.jl
+++ b/src/attributes.jl
@@ -218,3 +218,35 @@ struct SourceSolution <: AbstractTransitionAttribute end
 The current bound to the objective of the node.
 """
 struct TransitionObjectiveValueBound <: AbstractNodeAttribute end
+
+"""
+    RandomTransition <: AbstractNodeAttribute
+
+return a randomly selected transition from node `node`
+### Examples
+
+```julia
+get(model, RandomTransition(), node)
+```
+"""
+struct RandomTransition <: AbstractNodeAttribute end
+function get(sp::AbstractStochasticProgram, tr::RandomTransition, node::Int)
+    r = rand()
+    cdf = 0.
+    for tr in get(sp,OutTransitions(),node)
+        if cdf >= r
+             cdf += get(sp, Probability(), tr)
+        else
+            return tr
+        end
+    end
+    error("sum of probability $cdf < 1")
+end
+
+struct IsLeaf <: AbstractNodeAttribute end
+function get(sp::AbstractStochasticProgram, node::Int)
+    return isempty(get(sp, OutTransitions(), node))
+end
+function is_empty(sp::AbstractStochasticProgram, node::Int)
+    return get(sp, IsLeaf(), node)
+end