This is an analysis of the attempts to colonize at least 208 species of parasites and predators on about 75 species of pest insects in the field in Canada. There was colonization by about 10% of the species that were introduced in totals of under 5,000 individuals, 40% of those introduced in totals of between 5,000 and 31,200, and 78% of those introduced in totals of over 31,200. Indications exist that initial colonizations may be favoured by large releases and by selection of release sites that are semi-isolated and not ecologically complex but that colonizations are hindered when the target species differs taxonomically from the species from which introduced agents originated and when the release site lacks factors needed for introduced agents to survive or when it is subject to potentially-avoidable physical disruptions. There was no evidence that the probability of colonization was increased when the numbers of individuals released were increased by laboratory propagation. About 10% of the attempts were successful from the economic viewpoint. Successes may be overestimated if the influence of causes of coincidental, actual, or supposed changes in pest abundance are overlooked. Most of the successes were by two or more kinds of agents of which at least one attacked species additional to the target pests. Unplanned consequences of colonization have not been sufficiently harmful to warrant precautions to the extent advocated by Turnbull and Chant but are sufficiently potentially dangerous to warrant the restriction of all colonization attempts to biological control experts. It is concluded that most failures were caused by inadequate procedures, rather than by any weaknesses inherent in the method, that those inadequacies can be avoided in the future, and therefore that biological control of pest insects has much unrealized potential for use in Canada.