In a country such as Pakistan, where there is no vital registration system, estimating mortality levels and trends can be difficult. One way we can learn about mortality is to use indirect estimation techniques on census age distributions. This paper applies some of these techniques and evaluates the quality of the census data in the process. Various researchers have found evidence of a mortality pattern unique to South Asian populations, (Heligman 1985). They find that the expectation of life for females is less than that for males where values of e(O) range between 40 and 50 years, but that the opposite is true where e(O) is between 60 and 70 years. This suggests that as South Asian mortality declines, the sex differential disappears, and the pattern is more like that found in the rest of the world. Consistently high sex ratios above age 10 are characteristic of Pakistan (see Table 1); the sex ratio at birth varies but tends to stay above 105, the expected value for most .populations (Visaria 1971). The underenumeration of females at marriageable ages is the explanation given by Krotki (1985) for the high sex ratios at ages 10-14 and 15-19. The age distribution implies greater survival to those ages than one would expect in a high fertility, high mortality population (see Table 2).