I study the effects of service offshoring on white-collar employment, using highly disaggregated occupational data for the U.S.. I present a structural model of the firm's behavior that allows tractable derivation of labor demand elasticities for highly detailed occupations. I estimate the model using Quasi-Maximum Likelihood, to simultaneously account for the high degree of censoring of the employment variable and the small cross-sectional dimension of the panel. I find that service offshoring is skill-biased, because it raises employment among high-skilled occupations and lowers employment among medium- and low-skilled ones. Within each skill group, service offshoring penalizes tradeable occupations and tends to benefit complex non tradeable jobs.