Eighty-two periodontal patients were treated in a split mouth design with coronal scaling (CS), root planing (RP), modified Widman surgery (MW), and flap with osseous resection surgery (FO) which were randomly assigned to various quadrants in the dentition. Therapy was performed in 3 phases: non-surgical, surgical, and supportive periodontal treatment (SPT) < or = 7 years. Clinical data consisted of probing depth (PD), clinical attachment level (CAL), gingival recession (REC), bleeding on probing (BOP), suppuration (SUP), and supragingival plaque (PL). Because of the necessity to exit many CS treated sites due to breakdown, data for CS were reported only up to 2 years. All therapies produced mean PD reduction with FO > MW > RP > CS following the surgical phase for all probing depth severities. By the end of year 2 there were no differences between the therapies in the 1 to 4 mm sites. There were no differences in PD reduction between MW and RP treated sites by the end of year 3 in the 5 to 6 mm sites and by the end of year 5 in the > or = 7 mm sites. FO produced greater PD reduction in > or = 5 mm sites through year 7 of SPT. Following the surgical phase, FO produced a mean CAL loss and CS and RP produced a slight gain in 1-4 mm sites. RP and MW produced a greater gain of CAL than CS and FO following the surgical phase in 5 to 6 mm sites, but the magnitude of difference decreased during SPT. Similar CAL gains were produced by RP, MW, and FO in sites > or = 7 mm. These gains were greater than that produced by CS and were sustained during SPT. Recession was produced with FO > MW > RP > CS. This relationship was maintained throughout SPT. The prevalences of BOP, SUP, and PL were greatly reduced throughout the study and were comparable between sites treated by RP, MW, and FO while the CS sites had more BOP and SUP.