The study focused on the investigation of serious crimes against unwilling victims, thus excluding vice, narcotics, gambling, or traffic offenses. Information on current practices was obtained from a national survey of all municipal or county police agencies that employed more than 150 officers or that served jurisdictions with a 1970 population in excess of 100,000. Interviews and observations were conducted in more than 25 departments selected to represent different investigative styles. Data on investigation outcomes were obtained from the Uniform Crime Report tapes, samples of completed cases, and internal evaluations or statistics compiled by individual departments. Data on the allocation of investigation efforts was obtained from a computerized workload file maintained by the Kansas City Police Department. The findings show more about what does not work than what does. The study demonstrated that investigative activities play only a minor role in contributing to overall arrest rates, and that much of an investigator's time is consumed with administrative paperwork or attempting to locate and interview witnesses in cases that empirical evidence indicates have a small probability of being solved. Recommendations are as follows: (1) postarrest investigations should be coordinated more directly with the prosecutor; (2) patrol officers should be given a larger role in preliminary investigations; (3) additional resources should be devoted to processing latent prints; and (4) departments should distinguish between those cases which involve only routine clerical processing and those which require special investigation or legal skills. Reform impacts of the study are examined. Footnotes and 10 references are provided.