The conventional approach to modeling cancer incidence rates uses the direct and indirect methods of standardization to estimate the standardized risk ratio of exposure for cancer registry data. However, this method cannot deal with several confounding factors. An alternative approach is to apply a Poisson regression model to estimate the incidence rate ratio for counts of events from a cancer registry data. The exponential of coefficient of the Poisson regression model represents the risk ratio. The author applied the Poisson regression model to the cancer registry data from the north of Iran to estimate the age-sex incidence rate ratio using GLIM software. It was shown how one could use the model for counts of events that occurred in a population or a person-year of time. For this analysis with GLIM software, the numerator and denominator of rate are needed from each cell of a contingency table. This model is quite flexible for count data and can control for several confounding factors while detecting the factor which modifies the effect of exposure.