We measure human capital using the self-reported skill sets of nearly 9 million U.S. college graduates from professional profiles on LinkedIn. We aggregate skill strings into 48 clusters of general, specific, and managerial skills. Multidimensional skills can account for several important labor-market patterns. First, the number and composition of skills are systematically related to measures of human-capital investments such as education and work experience. The number of skills increases with experience, and the average age-skill profile closely resembles the well-established concave age-earnings profile. Second, workers who report more skills, especially specific and managerial ones, hold higher paid jobs. Skill differences account for more earnings variation than detailed measures of education and experience. Third, women show slower skill growth than men around typical ages of first motherhood. Gender differences in skills rationalize a substantial proportion of the gender gap in job-based earnings