Studies using animal models have been unable to determine the mechanical stimuli that most influence muscle architectural adaptation. We examined the influence of contraction mode on muscle architectural change in humans, while also describing the time course of its adaptation through training and detraining. Twenty-one men and women performed slow-speed (30^o·s^-1) concentric- (CON) or eccentric-only (ECC) isokinetic knee extensor training for 10 weeks before completing a 3-month detraining period. Fascicle length of vastus lateralis (VL), measured by ultrasonography, increased similarly in both groups after 5 wk (CON = +6.3±3.0%, ECC = +3.1±1.6%; mean = +4.7±1.7, p<0.05). No further increase was found at 10 wk, although a small increase (mean ~2.5%; n.s.) was evident after detraining. Fascicle angle increased in both groups at 5 (CON = +11.1±4.0%, ECC = +11.9±5.4%; mean = 11.5±3.2%, p<0.05) and 10 weeks (CON = +13.3±3.0%, ECC = +21.4±6.9%; mean = 17.9±3.7%, p<0.01) in VL only, and remained above baseline after detraining (mean = 13.2%); smaller changes in VM did not reach significance. The similar increase in fascicle length observed between the training groups mitigates against contraction mode being the predominant stimulus. Our data are also strongly indicative of: 1) a close association between VL fascicle length and shifts in the torque-angle relationship through training and detraining, and 2) that changes in fascicle angle are driven by space constraints in the hypertrophying muscle. Thus, muscle architectural adaptations occur rapidly in response to resistance training, but are strongly influenced by factors other than contraction mode.