This study provides the first systematic examination of patterns of occupational sex segregation in less developed countries. I apply log-linear and log-multiplicative models to six-category occupational data from 44 nations to describe and explain the segregation patterns in both industrialized and less developed countries. These analyses indicate that segregation patterns are far more variable in less developed countries than in developed nations. In most countries, women are indeed over-represented in professional, clerical, sales and service work, and men are over-represented in managerial and production occupations. However, this dominant pattern holds for only 74 percent of industrialized nations and merely 52 percent of developing countries. I identify four additional "variant patterns" of segregation that deviate from the dominant pattern in striking ways. Also, my findings cast doubt on the usefulness of index measures of sex segregation (such as D, the index of dissimilarity), because a model implying that patterns of segregation vary only in degree and not in shape can be rejected. I can explain 73.2 percent of the variability in segregation patterns with such covariates as relative economic development, fertility, the size of the service sector, labor force growth, and average female human capital. The effects of these variables in less-developed countries are often inconsistent with previous results for developed countries, suggesting that conventional explanatory theories of sex segregation require modification and elaboration. Of course, my results are based on highly aggregate occupational data, and future explanatory studies of occupational sex segregation in less developed countries would most likely benefit from using more disaggregated data.