Purpose: Using the visual world paradigm with the eye-tracking technique, this study examined the extent to which lexical tone similarity influences spoken word recognition. Method: In two experiments, participants were audibly presented with a target word and visually presented with the same target word, a tonal competitor, and two distractors, and they were required to identify the target word. In Experiment 1, the two tonal competitors shared either acoustically highly similar tones (e.g., target word: “阳台” / yang 2 tai2 /, “ balcony ” vs. competitor: “养子” / yang 3 zi3 /, “ adopted son ”) or acoustically lowly similar tones (e.g., target word: “阳台” / yang 2 tai2 /, “ balcony ” vs. competitor: “样本” / yang 4 ben3 /, “ sample ”). In Experiment 2, the acoustic similarity of the target words and the tonal competitors shared either acoustically highly similar tones or acoustically lowly similar tones or identical tones (e.g., target word: “阳台” / yang 2 tai2 /, “ balcony ” vs. competitor: “ 羊毛” / yang 2 mao2 /, “ wool ”). Results: The results of the two experiments consistently demonstrated a graded tonal competitor effect, in which acoustically highly similar tonal competitors attracted more visual attention than acoustically lowly similar tonal competitors. Conclusion: Tonal similarity plays a graded constraining role in spoken word recognition in Mandarin Chinese.