This dissertation presents a non-experimental, quantitative impact evaluation of PensiÛn para Adultos Mayores (PAM), the largest Mexican cash transfer program targeted at the elderly. Although an evaluation framework was part of the design of the PAM program, its rapid expansion rendered the evaluation framework useless after the Örst years. Using the Socioeconomic Conditions Section (MCS) of the Mexican National Income and Expenditure Survey (ENIGH) from 1992 to 2014 and the eligibility requirements of the PAM program at each expansion phase, I estimate intention-to-treat short-term e§ects on poverty, food insecurity, access to health services, and employment for eligible households/individuals in urban areas. Furthermore, I estimate intention-to-treat e§ects on poverty for the rural elderly and test if these e§ects have been constant or if they have been fading away over time. Given the data availability, program design and its subsequent roll-out, my identiÖcation strategy consists of using di§erence-in-di§erences (DD) and di§erence-in-di§erences-in-di§erences (DDD) approaches to estimate the e§ects. A Önal meta-analysis of the estimated impacts is presented for each outcome variable. Results show statistically signiÖcant and economically large constant impacts in the reduction of extreme and moderate poverty for the elderly living in rural areas. They also show smaller, but statistically signiÖcant short-term reductions in extreme poverty and moderate poverty for the eligible groups in urban areas. The estimated impacts on food insecurity were also signiÖcant, but only for less populated localities. Furthermore, the probability of having a job decreased for eligible male residents of households with only senior residents, for adult male residents aged 18 to 59, and for young male residents aged 15 to 20 living in households with at least one qualifying resident. The estimated impacts on access to health services were not statistically signiÖcant. Lastly, I estimated the budgetary needs of the program until 2051 to assess its Öscal viability.