A lively debate exists over the labor market returns to computer skills with findings ranging from large positive effects on earnings to the concern that estimated “pencil effects” are just as large. This paper provides the first evidence on the question from a randomized controlled trial providing computers to entering college students. We matched confidential administrative earnings data from California UI records to all study participants for seven years after the provision of computers. The experiment does not provide any evidence that computer skills have short- or medium-run effects on earnings. These null effects are found along both the extensive and intensive margins of earnings. In addition, the absence of positive labor market returns to computer skills does not appear to be due to positive effects on college enrollment resulting in delayed entry into the labor market. A non-experimental analysis of CPS data reveals large, positive and statistically significant relationships between home computers and labor market outcomes, which raises concerns about selection bias in non-experimental studies.