Although self‐talk during competitive sports is common and predictive of athletes' motivation, experiences, and performance, it is difficult to accurately assess self‐talk. An important, yet underexplored, next step in the assessment of self‐talk is to rely on a multi‐method approach. The present study sought to examine whether tennis players' self‐talk assessed either via self‐reports or via a live‐recorded procedure would relate to each other. Competitive tennis players (N = 120; Mage = 25.22; SDage = 9.82) were asked to perform multiple tennis exercises while verbalizing their thoughts, which were audio‐recorded and subsequently coded. Prior to exercise engagement, they indicated their fear of failure, while, after exercise engagement, they reported on their experienced pressure and self‐talk using questionnaires. There was substantial correspondence between the coded and self‐reported measure, allowing the estimation of a latent factor representing a multi‐method assessment of self‐talk. Moreover, in a theory‐consistent way, both latent factors representing negative and positive self‐talk were related to a hypothesized antecedent (ie, fear of failure), with negative self‐talk also relating to a hypothesized consequence (ie, perceived pressure). Overall, the present study shows that athletes' self‐talk can be measured reliably through different methods. Guidelines for the assessment of self‐talk in future research are provided.