OBJECTIVE To establish the extent of alcohol involvement in motor vehicle crashes occurring in Eldoret in western Kenya. METHODS A descriptive hospital-based study. METHODS Crash-involved patients aged 16 years and above presenting for treatment in all hospitals located in Eldoret town over a period of six months. METHODS Casualties were enrolled consecutively. A questionnaire eliciting demographics and crash circumstances was administered. Blood alcohol concentration (BAC) was evaluated either by breath tests or venous blood sample analysis in consenting casualties presenting within 10 hours of the crash. BAC levels of 5 mg% and greater were taken as a positive test; patients registering BAC levels equal to or greater than 50 mg% were considered as being intoxicated. RESULTS Of the 188 patients evaluated, 23.4% were BAC positive and 12.2% were intoxicated. Males were twice as likely as females to have been drinking prior to the crash (26.4% versus 13.6%; p = 0.08). Significantly greater proportions of night-time and weekend crashes involved intoxicated subjects (p = 0.02 and p = 0.03, respectively). Motor vehicle drives were the most affected by alcohol (60%), whereas pedestrians (33.3%), passengers (16%) and pedal cyclists (8.3%) were involved to a lesser extent. In comparison to passengers, drivers were eight times more likely to have been drinking (OR = 7.9, p = 0.04). CONCLUSIONS Alcohol is a contributing factor in a substantial proportion of traffic crashes occurring in western Kenya. Policy response and specific interventions for discouraging driving under the influence of alcohol, including the establishment and enforcement of a legal BAC limit are needed.