The aim of this paper is to update and critically analyze the public health relevance of available evidence about the causal relationship between cannabis use and psychosis or depression. There are conflicting views about this causal relationship. Two systematic reviews of existing evidence assessed this association and were published in 2004, but they came to different conclusions. From a public health perspective a thorough discussion is warranted before attributing any observed effect to potential biases. First, the impact of residual confounding in this causal relationship is discussed. We consider that the attenuation of estimates after controlling for confounding factors cannot be interpreted as evidence to support the claim that residual confounding fully explains this link. Secondly, taking into account the results of recent studies, the time-sequence and dose-response criteria of causality are discussed. The fact that unreported or subclinical psychological problems might have preceded and precipitated cannabis use is a very unlikely explanation when the cannabis-psychosis outcome link is assessed from different longitudinal studies. And finally, available evidence is interpreted with a broad view of public health and by taking into account the precautionary principle. We therefore provide reasons to support the case that stronger preventive actions against cannabis are still required in order to avoid the potential increased incidence of psychosocial health problems in the future.