The current study compared the effects of concentric-based (CONC), eccentric-based (ECC) and traditional concentric-eccentric (TRAD) resistance training on muscle strength, mass and architecture and the post-detraining retention of the training-induced effects in women. Sixty women were randomly assigned to unilateral volume-equated CONC, ECC or TRAD knee-extension training or control (N=15 per group). Before, after an 8-week intervention and 8-week detraining period, isokinetic concentric, eccentric and isometric torque were measured. Additionally, thigh lean mass was assessed by dual X-ray absorptiometry and vastus lateralis thickness, pennation angle and fascicle length by ultrasound. After training, concentric and isometric torque increased (p<0.05) similarly in all groups, while eccentric torque increased more in ECC than CONC [+13.1%, effect size (ES): 0.71(0.04/1.38)] and TRAD [+12.6%, ES: 0.60(0.12/1.08)]. Thigh lean mass increased in ECC [+6.1%, ES: 0.47(0.27/0.67)] and TRAD [+3.1%, ES: 0.33(0.01/0.65)]. Vastus lateralis thickness and pennation angle increased (p<0.05) similarly in all groups, while fascicle elongation was visible in ECC [+9.7%, ES: 0.92(0.14/1.65)] and TRAD [+7.1%, ES: 0.64(0.03/1.25)]. After detraining, all groups retained (p<0.05) similarly concentric torque. ECC and TRAD preserved eccentric torque (p<0.05), but ECC more than TRAD [+17.9%, ES: 0.61(0.21/1.21)]. All groups preserved isometric torque (p<0.05), but ECC more than CONC [+14.2%, ES: 0.71(0.04/1.38)] and TRAD [+13.8%, ES: 0.65(0.10/1.20)]. Thigh lean mass and vastus lateralis fascicle length were retained only in ECC (p<0.05), pennation angle was preserved in all groups (p<0.05) and thickness was retained in CONC and ECC (p<0.05). Including the eccentric phase in resistance training is essential to preserve adaptations after detraining.