Although available evidence indicates that vouchers improve service utilization among the target populations, we do not know whether increased utilization results from improved access (new clients who would not have used services without the voucher) or from shifting clients from non-accredited to contracted service providers. This paper examines whether the safe motherhood voucher program in Kenya is associated with improved access to health facility delivery using information on births within two years preceding the survey in voucher and comparison sites. Data were collected in 2010–2011 and in 2012 among 2933 and 3094 women aged 15–49 years reporting 962 and 1494 births within two years before the respective surveys. Analysis entails cross-tabulations and estimation of multilevel random-intercept logit models. The results show that the proportion of births occurring at home declined by more than 10 percentage points while the proportion of births delivered in health facilities increased by a similar margin over time in voucher sites. The increase in facility-based births occurred in both public and private health facilities. There was also a significant increase in the likelihood of facility-based delivery (odds ratios [OR]: 2.04; 95% confidence interval [CI]: 1.40–2.98 in the 2006 voucher arm; OR: 1.72; 95% CI: 1.22–2.43 in the 2010–2011 voucher arm) in voucher sites over time. In contrast, there were no significant changes in the likelihood of facility-based delivery in the comparison arm over time. These findings suggest that the voucher program contributed to improved access to institutional delivery by shifting births from home to health facilities. However, available evidence from qualitative data shows that some women who purchased the vouchers did not use them because of high transportation costs to accredited facilities. The implication is that substantial improvements in service uptake could be achieved if the program subsidized transportation costs as well.