In garden path (GP) jokes, a first dominant interpretation is detected as incoherent and subsequently substituted by a hidden joke interpretation. Two important factors for the processing of GP jokes are salience of the initial interpretation and accessibility of the hidden interpretation. Both factors are assumed to be affected by contextual embedding. We investigated this contextual impact with a priming manipulation in a two self-paced reading experiments (Ns = 45/44). Words that were semantically related to the initial interpretation, to the hidden interpretation, or nonrelated to both served as priming cues before the joke presentation. Only priming semantic content of the hidden interpretation reduced reading times of the final word of the GP joke compared with both other conditions, indicating facilitated detection and revision of the otherwise incoherent discourse. The results highlighted the impact of subtle contextual cues affecting the reinterpretation stage in GP joke comprehension.