Evidence has repeatedly shown that disparities in crime sentences can be attributed to certain variables considered outside the legal dimensions of the case. The majority of research that investigates factors that contribute to such disparities has primarily focused on crimes of varying severities adjudicated in the U.S. court system. We expand research on this topic by focusing on disparities in homicide sentences using data from over 9000 homicide cases tried in Colombia from 1980 - 2000. We specifically explore whether judges use substantive rationality when deciding the length of the offender´s sentence and if the sentence should be above the legal minimum set for the severity of the crime according to the criminal code under which it is adjudicated. Results reveal that disparities in homicide sentences can be attributed to extra-legal variables such as: the city in which the homicide trial took place, where the body of the victim was retrieved, and whether the defendant was identified by an ID parade. However, we also find evidence that suggests that legal variables such as the defendant´s previous criminal record and the aggravating circumstances of the case engender greater differences in sentence outcomes than non-legal variables previously mentioned. Explanations and policy implications are discussed.