Abstract Marine mammal interactions with fisheries create conflicts that can threaten human safety, economic interests and marine mammal survival. A deterrent that capitalizes on learning mechanisms, like fear conditioning, may enhance success while simultaneously balancing welfare concerns and reduce noise pollution. During fear conditioning, individuals learn the cues that precede the dangerous stimuli, and respond by avoiding the painful situations. We tested the efficacy of fear conditioning using acoustic stimuli for reducing California sea lion Zalophus californianus interactions from two fishing contexts in California, USA; bait barges and recreational fishing vessels. We performed conditioning trials on 24 individual sea lions interacting with bait barges. We tested for acquisition of conditioned fear by pairing a neutral tone with a startle stimulus. Avoidance was strongest in response to the startle stimulus alone, but low when paired with a neutral tone. From actively fishing vessels, we tested for fear conditioning by exposing sea lions to a neutral tone followed by a startle pulse, a startle pulse alone or a no sound control. We conducted playbacks from 146 (including 48 no sound control) stops over two summer fishing seasons (2013, 2014). The startle stimulus decreased surfacing frequency, reduced bait foraging and increased surfacing distance from the vessel while the conditioned stimulus only caused a mild reduction in surfacing frequency with no other behavioral change. Exposing animals to a pair of a conditioned stimulus with a startle pulse did not achieve the intended management outcome. Rather, it generated evidence (in two study contexts) of immediate learning that led to the reduction of the unconditioned response. Taken together, our results suggest that for fear conditioning to be applied as a non‐lethal deterrent, careful consideration has to be given to individual behavior, the unconditioned/conditioned responses and the overall management goals.