Conservation agriculture (CA) is an important aspect of the sustainable management of soybean [Glycine max (L.) Merrill] in production systems. This work evaluated the effects of soil management systems (SMS) combined with liming performed in 1986 and in 2008 on soil physical properties and soybean yield. The study used data from the 2008/09 until the 2015/16 soybean crop seasons from a long-term experiment settled in 1986 in Passo Fundo, southern Brazil. A split-plot design was used with the SMS as the main plots arranged in randomized blocks, and the cropping systems as the sub-plots randomized within SMS. The experiment evaluated the effects of four annually performed SMS: no-tillage (NT) and reduced-tillage (RT) (i.e. two CA systems), disk ploughing + disk harrowing (DPD) and moldboard ploughing + disk harrowing (MPD) [i.e. two conventional tillage (CT) systems]. Soil bulk density (rho(s)) and total, micro and macro porosities (phi(total), phi(micro) and phi(macro)) were evaluated in 2008 and in 2016 in the 0-2.5 cm (L0 - 2.5) and 10-12.5 cm (L10 - 12.5) soil layers. After 22 years of the beginning of the experiment and compared with the undisturbed soil (from a native forest area near to the experiment), all SMS increased rho(s) by an average of 28% in the L0 - 2.5 and 25% in the L10 - 12.5, decreased phi(total) by 13% (except for NT) in the L0 - 2.5 and 17% in the L10 - 12.5, and decreased phi(macro) by 36% in the L10 - 12.5; moreover the CT systems decreased phi(micro) by 8.7% and 8.1% in the L0 - 2.5 and L10 - 12.5, respectively. At the end of this 22-year period, in the L0 - 2.5, rho(s) increased by 6% from NT to RT, and by 11% from NT to CT systems; phi(total) decreased by 4%, and phi(micro) increased by 9% from NT to the other SMS. From 2008 to 2016, liming combined with each SMS modified soil properties in distinct manners: the combination increased rho(s) by 6% in NT (L0 - 2.5) and 5% in MPD (L10 - 12.5), decreased phi(total) by 4% in MPD (L10 - 12.5), increased phi(micro) in all SMS by an average of 6% (L10 - 12.5), and reduced phi(macro) by an average of 24% in CT systems (L10 - 12.5). Soybean yield was more variable as function of growing season (average from 1866 to 4449 kg ha(- 1)) as compared to SMS treatment [average from 3088 kg ha(- 1) (DPD) to 3276 kg ha(- 1) (NT)]. Considering a global analysis of the eight crop seasons, soybean yield in NT was on average 6% greater than that of DPD, but NT grain yield was similar to RT and MPD systems. No-tillage favored soybean yield in higher yielding environments, while DPD had the greatest soybean yield and adaptability in lower yielding environments. These findings suggest that the NT system outperformed the other SMS by providing greater or similar soybean yields and being the least harmful to soil physical quality as compared to the undisturbed soil.