Governments worldwide administer targeted social programs to improve the well-being of vulnerable groups, yet many eligible citizens do not take up these programs. This paper examines take-up of an unconditional cash transfer program for poor widows and divorcees in Delhi, India. Despite the considerable benefits, only one-third of eligible citizens are enrolled, with lower enrollment among more vulnerable women. I conduct a field experiment with over 1,200 pension-eligible women to identify barriers to program take-up and their distributive implications. One group of women is provided with only information about the program. Others receive information plus mediation: assistance with filling out the application form (basic mediation) or assistance engaging with political authorities (intensive mediation). I find that information alone raises application rates only among literate women. On the other hand, basic and intensive mediation increase average application rates by 41% and 70%, respectively. Furthermore, providing mediation changes the applicant pool to include more vulnerable women: those who are illiterate, politically disconnected, or lack autonomy in their household. While conventional wisdom suggests that application ordeals ensure take-up by those with the highest marginal utility of enrollment, I show that ordeals can interact with capabilities of poor citizens to select out those with a high need for the program. Simpler enrollment procedures and strengthened channels of bureaucratic mediation may facilitate more widespread and inclusive take-up.