The Spatial-Numerical Association of Response Codes (SNARC) effect refers to the phenomenon of faster left-hand responses to smaller numbers and faster right-hand responses to larger ones. The current study examined the possible long-lasting effects of magnitude-relevant stimulus-response compatibility (SRC) practices on the SNARC effect in a transfer paradigm. Participants performed a magnitude classification task including either SNARC-compatible or SNARC-incompatible trials as practice. They performed a parity judgment task in the subsequent transfer session, administered five minutes, one day or a week after the practice session. Results revealed significant SNARC effects after compatible practices and significant reverse SNARC effects after incompatible practices in all time-interval conditions. However, a control group without practice showed no reliable SNARC effect. These findings suggest that the SNARC effect can be influenced by magnitude-relevant associations formed a week previously, highlighting the long-lasting effects of magnitude-relevant SRC practices on the SNARC effect.