Abstract : As part of an effort to provide improved techniques and methods for intelligence collection planning and intelligence analysis, an experiment was conducted to assess the effectiveness of traditional tactical intelligence indications for the analysis of conventional military operations. Forty-six captains in the Intelligence Officers Advanced Course each assumed the role of staff officer in a G-2 section of an Infantry Division conducting a mobile defense in north-central West Germany. Each officer estimated the probability that each of 49 separate indications listed in Field Manual 30-5, Combat Intelligence, would occur, given the aggressor's known course of action. Each indication was evaluated with four separate courses of action--Attack, Defend, Delay, and Withdraw. Eleven indications were evaluated twice with each course of action to provide an estimate of reliability. The estimates made by individual officers were highly reliable. However, the variability in the estimates made by different officers for the same indication was extremely high, with an average range of estimates greater than .7 on a 0 to 1.0 scale. The logic underlying clusters of related indications could not be clearly identified for any of the four courses of action. The findings reveal that current indications of conventional military operations are either poorly understood or intrinsically inadequate for use in contemporary intelligence operations, or both.