Abstract Prey species assess predation risk by using either direct and indirect cues and both may contribute to a proper evaluation of the actual risk. As postulated by the risk assessment hypothesis, conspecific density may also provide useful information for tuning defensive responses. We tested this hypothesis using a combination of five density levels (1, 2, 4, 8 and 16 individuals) of Italian agile frog Rana latastei tadpoles and three treatments (control, predatory cues of common backswimmer Notonecta glauca and a waterjet of tap water as mechanical disturbance). Tadpole activity decreased in response to all stimuli but, as expected, backswimmer cues induced a stronger and lasting response. However, tadpole activity level did not vary with group size, thus providing no support for the risk assessment hypothesis and confirming that conspecific density might have less consistent effects on short-term behavioural responses than morphological and life history traits.