I need to fabricate some medical data. I must have 30 or so measurements on a scale of 0-1000. The results should be random within certain limits.

What are some common pitfalls when imitating random results?