Color imagery refers to the experience of color based on memories without its perception. Compared to color perception, it is not well known how color imagery is represented in the brain. Last year, we reported that we were able to decode recalled color imagery using signals obtained from multichannel subdural electrocorticography (ECoG) using pattern classification methods. To distinguish between decoding reflecting cue information that was used for color recall and decoding reflecting recalled color imagery, we trained a monkey to perform color recall tasks using two different cues for each color. In the task, a cue stimulus, either an achromatic scene image or a Fourier descriptor, was presented, and after a delay period, 1–2 colored shapes were presented sequentially as choice stimuli. To receive a reward, the monkey released a lever upon presentation of a correctly colored choice that was associated with the cue. The monkey was supposed to recall and maintain color imagery during the delay period. Using a power spectrum of ECoG signals during that period, we constructed support vector machine classifiers (decoders) to predict the monkey's color choice. To examine the generalization performance of the decoders across the tasks using different cues, we trained the decoders on the data set from the task using one type of cues (scene images) and tested them on the data set from the task using another type of cues (Fourier descriptors), and vice versa. Our results demonstrate that the decoders successfully predict the monkey's color choice regardless of the cue types, providing support that recalled color imagery can be decoded by our methods. We are now examining which brain areas and frequency bands are responsible for decoding color imagery.