There is a large and growing interest in non-consumptive effects (NCEs) of predators. Diverse and extensive evidence shows that predation risk directly influences prey traits, such as behaviour, morphology and physiology, which in turn, may cause a reduction in prey fitness components (i.e. growth rate, survival and reproduction). An intuitive expectation is that NCEs that reduce prey fitness will extend to alter population growth rate and therefore population size. However, our intensive literature search yielded only 10 studies that examined how predator-induced changes in prey traits translate to changes in prey population size. Further, the scant evidence for risk-induced changes on prey population size have been generated from studies that were performed in very controlled systems (mesocosm and laboratory), which do not have the complexity and feedbacks of natural settings. Thus, although likely that predation risk alone can alter prey population size, there is little direct empirical evidence that demonstrates that it does. There are also clear reasons that risk effects on population size may be much smaller than the responses on phenotype and fitness components that are typically measured, magnifying the need to show, rather than infer, effects on population size. Herein we break down the process of how predation risk influences prey population size into a chain of events (predation risk affects prey traits, which affect prey fitness components and population growth rate, which affect prey population size), and highlight the complexity of each transition. We illustrate how the outcomes of these transitions are not straightforward, and how environmental context strongly dictates the direction and magnitude of effects. Indeed, the high variance in prey responses is reflected in the variance of results reported in the few studies that have empirically quantified risk effects on population size. It is therefore a major challenge to predict population effects given the complexity of how environmental context interacts with predation risk and prey responses. We highlight the critical need to appreciate risk effects at each level in the chain of events, and that changes at one level cannot be assumed to translate into changes in the next because of the interplay between risk, prey responses, and the environment. The gaps in knowledge we illuminate underscore the need for more evidence to substantiate the claim that predation risk effects extend to prey population size. The lacunae we identify should inspire future studies on the impact of predation risk on population-level responses in free-living animals.