Two experiments (exp.) were conducted to determine the effect of dietary lysine restriction during the grower phase and subsequent dietary lysine concentration during realimentation on the performance, carcass characteristics and nitrogen (N) balance of growing–finishing pigs. Seventy-two pairs of crossbred [meatline sires × (Large White × Landrace) dams] pigs (36 pairs entire male, 36 pairs female) were used in a 2 × 4 factorial performance study (exp. 1, 35 to 96 kg). The two factors were Diet1 (two lysine levels: LO or HI, day (d) 0 to 35) and Diet2 (four lysine levels: LO, M1, M2 and HI, d36 to slaughter). A further 30 boars were used in a 2 × 3 factorial N-balance study (exp. 2, 31 to 42 kg; Diet1: LO or HI; Diet2: LO, M3 and HI). The crude protein (CP) and lysine contents of the five isoenergetic (14.0 MJ DE/kg) diets were: LO (139 g/kg CP, 8.1 g lysine/kg), M1 (162 g/kg CP, 9.7 g lysine/kg), M2 (190 g/kg CP, 11.4 g lysine/kg), M3 (180 g/kg CP, 10.7 g lysine/kg) and HI (217 g/kg CP, 13.1 g lysine/kg). Diets were based on barley, wheat and soyabean meal with added vitamins, minerals and amino acids. Ratios of amino acids relative to lysine were similar for all diets. From d0 to 35 in exp. 1, pigs fed LO had lower daily gains (ADG: 785 vs. 939 g/day; P < 0.001), daily lysine intakes (DLYIN: 14.4 vs. 22.6 g/day; P < 0.001) and poorer feed conversion ratios (FCR: 2.28 vs. 1.84 kg/kg; P < 0.001) than pigs fed HI. From d36 to slaughter, pigs initially fed LO had higher ADG than those initially fed HI (948 vs. 879 g/day; P < 0.01). However, from d0 to slaughter, pigs initially fed LO had lower ADG (868 g/day; P < 0.01) and DLYIN (19.1 g/day; P < 0.001) than those fed HI (913 g/day and 23.3 g/day respectively). There was a Diet1 × Diet2 interaction for FCR from d36 to slaughter and d0 to slaughter ( P < 0.05). FCR of pigs previously fed LO improved with each increase in dietary lysine concentration up to 13.1 g/kg (HI), but improvement was only evident for lysine concentrations up to 11.4 g/kg (M2) for pigs previously fed HI. In exp. 2, Diet1 was fed for an initial 5-day collection period (after adjustment to crate; C1). Diet2 was then fed and immediately there were 2 successive 5-day collection periods (C2 and C3). A Diet1 × Time interaction ( P < 0.001) indicated that although N retention (NRET) during C1 was lower for pigs fed LO, there was no difference during C2 and C3 in NRET of pigs initially fed either LO or HI (14.5 vs. 25.0 g/day; 23.2 vs. 23.0 g/day; 26.0 vs. 26.6 g/day, for C1, C2 and C3, respectively). However, a similar interaction indicates that although N excretion (NEX) was 23% lower for LO compared with HI pigs during C1, there was no residual reduction in NEX of pigs initially fed LO ( P < 0.01: 8.9 vs. 11.6 g/day; 12.3 vs. 12.7 g/day; 13.8 vs. 14.0 g/day, for C1, C2 and C3, respectively). Exp. 1 indicated improved performance and an ability to respond positively to higher lysine concentrations during realimentation for previously restricted compared with unrestricted pigs, however, overall performance was reduced. However, exp. 2 indicated that although NRET of initially restricted pigs increased during realimentation, it did not surpass that of unrestricted pigs. This apparent discrepancy may have been due to the time period of the N balance study.