Monitoring rare species of wild ungulates is critical for their conservation management. The Italian roe deer Capreolus capreolus italicus was recently confirmed to be a subspecies in Mediterranean habitats of central and southern Italy. We have monitored this subspecies at Castelporziano, near Rome, since 1988, and detected an abrupt population decline in 2001. We compared distance sampling surveys undertaken before and after the population crash to determine the ability of this method to detect variations in animal density and to investigate which factors may influence the precision of the estimates. We used radio tagged roe deer to evaluate the accuracy of our surveys, comparing distance sampling and mark-resight estimates at the same site, and studying the behavioural reaction of the animals to the presence of an observer. We found that before the crash distance sampling surveys attained a good precision but that the survey conditions influenced both precision and accuracy. Post-crash surveys were less precise, but the difference in density, before and after the crash was highly significant, indicating the potential of the method to quickly detect density variations and so to allow wildlife managers to react without delay to a crisis. The direct test of distance sampling assumptions showed that estimates were almost unbiased. We conclude that the methodology was successful for monitoring this population and that it may be used in other situations where ungulates are rare and efficient monitoring must be attained with restricted budgets.