We show that the distinction between job spells and employer spells matters for returns to tenure. Employer spells encompass between-job wage movements linked to promotions or demotions while job spells don't. Using a 1% sample of the British workforce over the period 1975-2010, we find that a significant proportion of the return to employer tenure arises due to job changes within employer spells. Conditional on tenure with employer, the return to job tenure is negative. This suggests that any positive effects of job-specific human capital on wage growth within jobs are outweighed by the effects of job changes within firms.