From the point of view of workplace safety, it is important to know whether having a temporary job has an effect on the severity of workplace accidents. We present an empirical analysis on the severity of workplace accidents by type of contract. We used micro data collected by the Italian national institute managing the mandatory insurance against work related accidents. We estimated linear models for a measure of the severity of the workplace accident. We controlled for time-invariant fixed effects at worker and firm levels to disentangle the impact of the type of contract from the spurious one induced by unobservables at worker and firm levels. We found that workers with a temporary contract, if subject to a workplace accident, were more likely to be confronted with severe injuries than permanent workers. When correcting the statistical analysis for injury under-reporting of temporary workers, we found that most of, but not all, the effect is driven by the under-reporting bias. The effect of temporary contracts on the injury severity survived the inclusion of worker and firm fixed effects and the correction for temporary workers' injury under-reporting. This however does not exclude the possibility that, within firms, the nature of the work may vary between different categories of workers. For example, temporary workers might be more likely to be assigned by the employer dangerous tasks because they might have less bargaining power. The findings will be of help in designing public policy effective in increasing temporary workers' safety at work and limiting their injury under-reporting.