PURPOSE: To improve operation feeling by computing operation characteristic at the time of accelerator operation, on the basis of a detection signal for an accelerator operation variable, and using a calculation result for learning and correcting target opening characteristic for throttle opening, depending on the accelerator operation variable.

CONSTITUTION: When acceleration status is identified at a step 16, an actual accelerator operation speed dsAim from a first acceleration operation variable SAim to an 'i'-th accelerator operation variable SAim is computed at a step 17. and a mean operation variable SAim for an accelerator operation variable SA during acceleration is computed at a step 18. Then, an operation speed dSA at the mean operation variable SAim is read out from an operation speed characteristic map at a step 19. Thereafter, a reference operation speed dSA and an actual accelerator operation speed dSAim are compared to give a speed ratio dSAy at a step 20. Then, the ratio dSAy is weight averaged for the previous learning correction value KLim-1, and a learning correction value KLim is obtained for updating the previous value and saved at a step 21. Thereafter, the value KLim is used to correct a target opening of throttle opening as opening corresponding to accelerator operation tendency.