Soil quality evaluation as a decision-making tool to improve understanding of soil quality is essential for grading croplands and adopting proper agricultural practices. Various methods of soil quality evaluation have been developed, which have occasionally generated inconsistent evaluation results between differing soil types. The applicability of these techniques is seldom tested before implementing an evaluation method on a specific soil region. Fluvisol is an important soil resource for agriculture in China, especially for irrigation districts along the lower Yellow River. In the present study, the soil quality of two typical agricultural counties (Yucheng and Kenli) along the lower Yellow River was evaluated using four commonly utilized methods. In the two counties, the overall spatial patterns of soil quality derived from the four methods were similar, with differences in details existing among these methods. The soil quality in Yucheng, ranging from moderate to high, is superior to that observed in Kenli, where salinity is the primary limiting factor. In addition, the applicability of soil quality evaluation methods on the Fluvisol was investigated. It was found that the integrated quality indexing-linear scoring (IQI–LS) and the Nemoro indexing-linear scoring (NQI–LS) methods were the most accurate and practical of the four methods studied. These methods, which are based on the total data set of indicators, show better performance for soil quality evaluation on a Fluvisol. Further, different evaluation methods based on the minimum data set of indicators were compared, considering both the accuracy of the evaluation and the economic cost of obtaining the soil data. The results from the present study indicate that the IQI–LS method based on the minimum data set of indictors is recommended for large-scale soil quality evaluations. •Four soil quality evaluation methods were compared for a Fluvisol.•IQI–LS and NQI–LS methods performed better in soil quality evaluations.•For large-scale studies, the IQI–LS based on a minimum data set is recommended.