A general assumption is that science is just organised commonsense. It is noted that translation involves a two-way pathway between basic laboratory science and patient care, and that some scientific findings have implications for prevention rather than treatment. A succinct critique follows on the key features that differentiate science and commonsense. The main part of the paper discusses six rather different examples of translation that went awry because people treated science and commonsense as equivalent. Examples based on empirical evidence of translation going awry include (i) the claim that only early intervention can bring lasting benefits; (ii) the claim that the main policy goal for children should be the elimination of all stresses; (iii) the claim that exposure in utero to maternal smoking causes ADHD and conduct disturbance; (iv) the claim that tax benefits should be used to encourage couples to marry; (v) the effects of profound institutional deprivation are similar to those of any adversity; and (vi) environmental effects are largely independent of genetic influences. Much of science is ‘unnatural’ in the sense that technical tools (such as imaging or DNA) are employed, or because animal models are used, or because unusual comparisons are made. Science cannot be based solely on an inductive process; rather, there must be some form of experiment and the testing of two or more alternative explanations. Translation needs to be based on top quality science and an appreciation that even the best science needs to take account of multiple strategies and multiple evaluations.