Ecotoxicological tests with plants are often used to help regulators setting soil screening values of many contaminants, arsenic (As) included. Yet, the measurement of a comprehensive set of endpoints in representative plant species is very time-consuming and hard to achieve in a single test. This study evaluated the best-suited endpoints to assess As toxicity and contamination effects in crop plants grown in oxidic soils, with a focus on Tropical agroecosystems. Our aim is to recommend an approach for setting soil screening values that uses arsenic ecotoxicological studies chosen on the basis of the sensitivity and reliability of endpoints in plant growth tests. From the measured endpoints, the most sensitive ones were: first germination count > relative leaf area > total dry mass > germination speed index, while the most reliable endpoints were: first germination count >total dry mass=germination speed index>plant height. The species Phaseolus vulgaris and Zea mays were the most and least sensitive to arsenic toxicity, respectively. The use of extractable As concentrations for toxicity characterization allows estimating more realistic soil screening values for arsenic in Tropical soils.