This paper identifies the variables related to road accidents, by means of nested Poisson and Negative Binomial models, for two urban zones. In zone 1, mixed-industrial land use predominates, while zone 2 is a larger area than includes zone 1 and has mainly mixed-residential land use. Zone 1 has just four road segments, zone 2 has 39 road segments.The best nested Poisson model was selected for each zone using the Stepwise Algorithm. Given that the road accidents data have over-dispersion, also Negative Binomial models were used.Results point out that, road accidents in zone 1 are related to BRT and buses, and inversely related to unitary freight trucks. In zone 2, road accidents are positive related with week of the day, rain, and big buses, and negatively related with articulated freight trucks. Week of the day has the highest estimated value, probably because traffic speeds are higher and traffic monitoring is lower on weekends. Results are useful to generate policies for reducing road accidents in urban mixed industrial zones. In order to improve estimations, other variables must be analyzed and others GLM must be proved.