If you're fitting a simple random effects model, consider using xtnbreg (see http://www.stata.com/manuals14/xtxtnbreg.pdf).

If you're building a multi-level model with random coefficients, consider using menbreg, which is available since version 13 (see http://www.stata.com/manuals14/memenbreg.pdf).

This book is detailed and good and has Stata  code throughout

http://www.cambridge.org/gb/academic/subjects/statistics-probability/statistical-theory-and-methods/negative-binomial-regression-2nd-edition