As one kind of simple deductive reasoning, the syllogism reasoning contains a premise which is made up of two nature judgments and a conclusion of one nature judgment. Previous studies of syllogism reasoning have mainly concentrated on how individual factors (such as belief bias, mode of thinking, etc.) would influence the syllogism reasoning while less emphasis has been laid on the external factors. In recent years, The influence of logic training on reasoning has attracted researchers’ attention, some researchers have previously found that logic training would not improve the scores of syllogism reasoning, while other researchers achieved completely different results. One potential reason for the controversial results may be the ignorance of individual differences in previous studies. Therefore, on the basis of existing research, the current study discusses the impact of logic training on the syllogism reasoning of college students with different rational thinking modes. A total of 105 college students (47 male and 58 female) were randomly selected to participate in the experiment. Before the experiment, each of them filled in a scale of Rational – Experiential Inventor (REI-40). Based on the scores of REI-40, we chose the highest 1/3 (N = 35) and lowest 1/3 (N = 35) as the high and low rational thinking groups respectively to conduct the experiment. After that, a 50-question syllogism reasoning test was carried out. The syllogism reasoning included two formats (AC-BC format and BC-AB format), and only the correctness of AC-BC format was examined. Then, the logic training was carried out, and participants were asked to do another 50-question syllogism reasoning test after the training. The results suggested that (1) The average response time of individual’s syllogism reasoning with high rational thinking mode is faster than those with low rational thinking mode; The average accuracy of non-conflicting task type problem’s syllogism reasoning is significantly higher than that of conflict task type problem; Logic training can improve the average response time and correction rate of individual’s syllogism reasoning; (2)The average accuracy of non-conflicting task type problem’s syllogism reasoning is still significantly higher than that of non-conflict task type problem after logic training; (3)Logic training can improve the average correction rate of individual’s conflict task type with low rational thinking mode. This means that logic training can significantly improve individual’s reasoning performance, but it can not completely eliminate the influence of belief bias on reasoning. Moreover, logic training has more obvious effect on improving individual’s reasoning performance with low rational thinking mode. In a word, this study suggests that there are individual differences in the impact of logic training on syllogism reasoning, which is helpful for further elucidating the impact of logic training on individual syllogism reasoning in different rational thinking modes. From the experimental results, it can be seen that logic training can weaken the influence of belief bias on reasoning, and effectively improve the reasoning performance of individuals with lower rational thinking mode, but the effect of improving individual rational thinking mode is not obvious. Further research can use different training methods to improve the reasoning performance of individuals with highly rational thinking patterns.