Sentinel plasticine prey has been increasingly used to estimate predation pressure under different experimental treatments or habitats. The use of plasticine prey, however, may bias the estimation of predation pressure, as they are designed to account for predation by organisms which visually recognize shapes and colors, but not behavioral or chemical characteristics of the prey. To evaluate the limitations of sentinel plasticine prey, we compared attack rates between real prey (dead and live mealworms) and plasticine molded into shape of caterpillars and cubes in a monsoonal tropical rainforest of south-eastern China. The highest attack rates were observed on dead prey followed by live prey and plasticine models. Additional comparisons of attack rates among plasticine models of different shapes in the three biomes showed inconsistent results, suggesting that predators across different habitats may not necessarily recognize the shapes of plasticine. The use of sentinel plasticine models is useful for ecological studies, but we recommend that both live and plasticine models should be used at least for the pilot studies to measure the potential bias. This is especially important when the plasticine models are compared among different habitats where attack rates on plasticine caterpillars may not represent relative differences in predation pressure.