Workers entering the labor market are uncertain about their skill set. Standard human capital theory assumes workers have perfect information about their skills. In this paper, I argue that skill uncertainty can explain one type of worker moves that standard human capital theory cannot: moves between jobs where they perform different kinds of tasks. I consider workers who have a multi-dimensional bundle of labor market skills and begin their careers uncertain about their skill levels. I construct a model that links learning about skills to the tasks performed in jobs: the more intensely a job uses a particular skill, the more the workers learn about their true level of that skill. The model also contains a skill accumulation motive: as workers use a skill they gain additional amounts of it. A simplified version of the model suggests that if skill uncertainty were the dominant force workers would switch between jobs that use skills in different ratios but similar total levels. On the other hand, if skill accumulation were the dominant force they would switch between jobs that use similar ratios of skills but higher total levels. Linking data on workers from the National Longitudinal Study of Youth 1979 with occupational characteristics from the US Department of Labor O* NET database, I show that worker mobility across different task mixes is common and I estimate the model parameters. The current results indicate that skill uncertainty explains approximately 30% of worker mobility across different task ratios.