In this study automatic processing of verbal information was investigated in 22 clinically depressed inpatients and 22 healthy controls in a longitudinal design. A semantic priming task with word pronunciation was administered twice, about 7 weeks apart. Following brief presentations of prime words, subjects had to read target words aloud as quickly as possible. Prime words were directly related, indirectly related, or unrelated to the target words. Stimulus onset asynchrony (SOA) of prime and target was 250 ms. In the course of inpatient treatment, patients recovered significantly. Semantic priming occurred in both study groups for the directly and indirectly related conditions across both testing times. As expected, directly related primes resulted in stronger priming than indirectly related primes. Patients and controls did not differ in semantic priming at either time or relatedness condition. Size of priming was not associated with depression severity, anxiety level, intelligence, medication, or clinical features. We conclude that depression is not characterized by dysfunctional automatic processing of verbal information.