The speech intelligibility of patients who have undergone partial surgical resection of the maxilla was investigated. The intelligibility characteristics of connected discourse in each patient were compared under three experimental conditions: presurgical, postsurgical with immediate prosthetic obturation, and postsurgical with definitive prosthetic obturation. The ability of untrained listeners to differentiate between presurgical speech and postsurgical speech following definitive prosthetic treatment was also studied. Each subject recorded a standard passage of connected discourse, the Rainbow Passage, under the three conditions stated above. Each of the 24 recordings was then segmented into 23 phrases and presented over earphones to listeners. A group of 10 listeners evaluated each recording, or a total of 240 listeners was used. The recordings were graded by the number of words correctly written. Analysis of variance showed no significant loss in intelligibility across the three conditions. A group of 50 listeners than evaluated sentence pairs containing samples of presurgical speech and speech produced following surgery and definitive prosthetic obturation. Their task was to judge whether there was a difference between the two speech samples. This was done in a classroom over loudspeakers. Sign-test procedures revealed that listeners could differentiate presurgical speech samples from postsurgical speech with definitive prosthetic obturation in four of the eight patients studied.