I would like to know if this is a good way to do this job, both from the point of view of elegance and efficiency. Maybe it needs to be more strict? I may create lists of 100,000 elements and don't want to run into space/time problems. I want to create an infinite list of Floats (for purposes of lazily zipping or writing later), with the following form: a random number of zeros, followed by a single random non-zero value, repeat the pattern. I call it a "break series" for reasons I won't go into here, but in practice, I want to choose the number of zeros in each section from a range of integers, and I want the non-zero values to have a particular frequency distribution. Here's some code I wrote: import System.Random import Control.Monad import Control.Monad.Random breakSeries :: Rand StdGen [Float] breakSeries = do x <- getRandomR (0.0::Float, 1.0) -- Hacking a simple frequency distribution of values of d. d <- if x < 0.6 then getRandomR (5.0::Float, 8.0) else if x < 0.8 then getRandomR (12.0::Float , 24.0) else getRandomR (60.0::Float , 90.0) n <- getRandomR (4::Int, 6) let bits = replicate n 0.0 ++ [d] liftM (bits++) breakSeries test = print . take 20 . evalRand breakSeries =<< newStdGen