Pesticides, by design, are biocides; their value lies in their ability to kill noxious or unwanted organisms. But they are rarely selective. Most act by interfering with fundamental biochemical and physiological processes that are common to a wide range of organisms - not only pests, but ourselves as well. The safest assumption is that every pesticide is harmful to all organisms until the contrary is proved. But establishing such proof is, inevitably, an impossible task. There are too many organisms to test and too many possibilities to explore, given reasonable constraints on time and cost. Pesticide development would become prohibitively expensive and farmers would be denied many potentially useful compounds. Thus in practice, assessing the hazards of pesticide use relies on a balance of evidence accumulated through several different lines of enquiry. Thresholds are set and codes of safe practice established on the basis of biochemical studies, tests on a limited number of organisms and monitoring of wildlife in field experiments. Inevitably, though, it is only after widespread use that many problems become apparent.