The ambiguity in the relationship between self-employment and educational attainment is well documented in the literature. Using an extensive individual level dataset from Periodic Labor Force Survey, we estimate the probability of being self-employed in India based on educational attainments. Our results suggest that the probability of being self-employed rises for an individual with education but not monotonically so. Indeed, the impact of education on likelihood of self-employment does not convey much information without considering how the effect varies across gender, caste, age, household size, religion, and industry as various cohorts chosen for this study using 418,297 observations. The probability to be self-employed varies considerably based on gender, caste and age when the level of education rises. A cohort based analysis for determination of self-employment is novel for India along with the findings where college educated women show higher probability of self-employment than men, for example. The importance of considering the non-linearity in the relationship between self-employment and education, usually part of analytical frameworks but inadequately addressed empirically, should be useful for better policies on the interaction between human capital and occupational choice. Robustness analysis considering further cohort effects in terms of household size and religion, buttresses our benchmark results.