To learn whether criticism and regulation of research practices have been followed by a reduction of deception or use of more acceptable approaches to deception, the contents of all 1969, 1978, 1986, and 1992 issues of the Journal of Personality and Social Psychology were examined. Deception research was coded according to type of (non)informing (e.g., false informing, consent to deception, no informing), possible harmfulness of deception employed (e.g., powerfulness of induction, morality of the behavior induced, privacy of behavior), method of deception (e.g., bogus device or role, false purpose of study, false feedback), and debriefing employed. Use of confederates has been partly replaced by uses of computers. "Consent" with false informing declined after 1969, then rose in 1992. Changes in the topics studied (e.g., attribution, socialization, personality) largely accounted for the decline in deception in 1978 and 1986. More attention needs to be given to ways of respecting subjects' autonomy, to appropriate debriefing and desensitizing, and to selecting the most valid and least objectionable deception methods.