The Kassel Stuttering Therapy (KST) is a computer-assisted German adaptation of fluency shaping. 400 clients (aged 9-65 years) completed the 2- to 3-week in-patient intensive treatment. Long-term data 1 year (n=238), 2 years (n=69), and 3 years later (n=69) are reported. Objective disfluency rates in 4 different speech situations (% stuttered syllables, %SS) were reduced from before to after intensive treatment from 12.6 to 1.6%SS and remained in the following 3 years between 3.2 and 3.8%SS. The disfluency changes from before therapy to 1 year later or more showed effect sizes of d > 1.1. Therapy effects were strongest for speech situations with the highest communicative demands (calling an unknown person by phone, interviewing passers-by on the street). Relapses were mostly temporary and most frequent within the first 6 months after intensive treatment. Disfluencies 1 year after the intensive course correlated highly with those 2 years (ρ=0.81) and 3 years (ρ=0.76) after. Subjective stuttering data (self-rating of stuttering severity and of avoidance of speech or speech situations) showed, somewhat damped, the same approximately L-shaped progress as the objective disfluencies. Speech rates and speech naturalness increased slightly. Covert assessment of disfluency by telephone with a subsample of older children yielded higher rates than overt assessment. Monetary compliance incentives increased the rate of individual home practice after intensive treatment. The presented method fulfils almost all criteria required from a stuttering therapy which may be called effective. Permanent freedom from self-monitoring could be achieved only in a minority of clients. The behavioural data of the therapy effects are supported by parallel functional brain imaging findings in a small sample of clients.