Go/No-Go (GNG) training has been shown to influence food evaluations, typically by reducing the attractiveness of no-go-associated foods. However, the cognitive mechanisms underlying these effects remain under debate, with both associative and inferential explanations proposed. The present preregistered study investigated whether cognitive load moderates the impact of GNG training on food evaluations, thereby testing the relative contribution of efficient versus effortful processes. Ninety-one female participants completed a GNG training task under either high or low cognitive load, with food stimuli from three categories assigned to go, no-go, or untrained conditions. Evaluations of no-go, go and untrained food stimuli were collected pre- and post-training, and awareness of the stimulus-response contingencies was assessed after training. Results revealed that cognitive load significantly reduced contingency awareness but did not affect GNG task accuracy or training-induced changes in food evaluation. GNG training led to more positive evaluations of go stimuli relative to no-go and untrained stimuli, indicating a go valuation effect. However, no-go stimuli were not devalued compared to untrained food stimuli. These findings suggest that the effects of GNG training on food evaluations may reflect efficient learning processes. These processes appear to be resilient to cognitive load manipulations, particularly when the training is conducted at the category level, with one stimulus category consistently paired with go and another with no-go. Overall, the findings provide insights into the design of GNG-based interventions and suggest that category-based training may be robust across varying levels of cognitive interference.