When PW's were taken, they were handed to the provost of the capturing division, and were interviewed as per orders:
GENERAL ORDERS, No. 72
HEADQUARTERS ARMY OF THE POTOMAC,
Washington, February 26, 1862.
All deserters from the enemy, prisoners, and other persons coming within our lines will be taken at once to the provost-marshal of the nearest division, who will examine them in presence of the division commander, or an officer of his staff designated for the purpose. This examination will only refer to such information as may affect the division and those near it, especially those remote from general headquarters.
As soon as this examination is completed--and it must be made as rapidly as possible---the person will be sent, under proper guard, to the provost-marshal-general, with a statement of his replies to the questions asked. Upon receiving him the provost-marshal-general will at once send him, with his statement, to the chief of staff of the Army of the Potomac, who will cause the necessary examination to be made. The provost-marshal-general will have the custody of all such persons. Division commanders will at once communicate to other division commanders all information thus obtained which affects them.
By command of Major-General McClellan:
S. WILLIAMS,
Assistant Adjutant-General.
Intelligence gathering was the province of the cavalry, and the provost-marshal. In McClellan's army this was BG Andrew Porter, who ran an efficient and fairly accurate int cell albeit with one bad apple - John C. Babcock. In September '62 Lincoln ordered Porter to Pennsylvania and McClellan fought the Maryland Campaign without parts of his intelligence staff. McClellan appointed BG Massena Patrick as P-M on 6th October '62, and he remained in post for the whole war.
John C. Babcock was the only member of McClellan's int to continue into Burnside's era, and he remained in post the whole war. This was unfortunate, as Babcock was the man generating overestimates of enemy strength. His orbat charts contained 36 infantry regiments that weren't with the enemy army as of June '62, and 46 by October '62. When estimating enemy strength, by mid-1862 two primary sources were in use:
1. Babcock's orbat charts
2. Numbers of rations issued (128,000, excluding Jackson on 21st June '62 for example, but only 59,000 by May '63)
The orbat charts were compiled from prisoner interviews. Prisoners were ask what regiment they were in, who their brigadier and major general was, and for any other details. Babcock's charts were very accurate, apart from 36+ infantry regiments reported present who had no known generals over them. The change Sharpe made was that he decided to keep these "maybes" out of the general estimate.