Facing the impacts of climate change on rice production in Taiwan, a new cultivation system named “the middle cropping season cultivation system” was proposed which rice is grown from May to October to prevent severe drought in the first cropping season (February to June) and unpredictable rainfall in the second cropping season (August to November). To evaluate the agronomic performance of diverse rice accessions grown in these seasons, it is inevitable to face genotype-by environment interaction (GEI) as GEI provides chance to find suitable accessions with wide adaptation or narrow adaptation to certain environments. Therefore, this study aims to evaluate GEI of traits related to heading, yield component and grain quality from 140 rice accessions through mutiple statistical tools in the first, middle and second cropping seasons in Taiwan. Genotype main effect plus GEI (GGE) biplot, additive main effect and multiplicative interaction (AMMI) model, a new quantitative genotypic stability measure named weighted average of absolute scores (WAAS) and Roemer’s environmental variance (Roemer’s EV) were applied to discuss GEI structure for days to heading (DTH), grain weight per plant (GW) and amylose content (AC) of the accessions. Generally, the results of winning accessions with both high mean value and wide adaptation from GGE biplot, AMMI1 biplot and WAAS were consistent. For DTH, GEI explained relatively small proportion of variance (6.87%) compared to G (54.6%) and E (37.4%). From the agronomic concept of stability, G95 was the most stable accession in both AMMI2 biplot and WAAS; however, G102 was the most stable one from the biological concept of stability by Roemer’s EV. For GW, GEI explained large proportion of variance (32.9%) compared to G (30.4%) and E (17.4%). GGE biplot showed that the best-performing accession in the middle cropping season was G109, which was totally different from that in the first and the second cropping seasons; however, G50 was considered as the best accession with high mean performance and wide adaptation through WAAS (WAASY), which was recommended to cultivate in both the first and the second cropping seasons. As for AC, GEI also explained a relatively small proportion (2.87%) with most of the variance explained by G (94.9%). G64 was selected as the best accession with AC close to our targeted accession G38 and high stability through WAASY from both agronimic and biololgical concept of stability. Our study indicated that the descriptive results from these statistical tools could support each other; nevertheless, the graphical tools should be used cautiously to prevent over-interpretation. Still, the selected well-performing accessions could provide information for further developing cultivars with wide adaptation or narrow adaptation under the threats of climate change.