Lagrange multiplier (LM) and Wald type tests for overdispersion and serial correlation using the generalized method of moment approach are proposed for a time series count data regression model. The small sample properties of the tests are compared by Monte Carlo experimentation. It is found that a traditional test for overdispersion has both better size and power properties than the LM and Wald tests when the sample size is small. The Wald and LM tests of serial correlation have both better size and power than traditional portmanteau tests.