Although brain computer interfaces (BCI) based on electroencephalographic (EEG) signals have been studied increasingly over decades, their performance is still limited in two main aspects. First, the shorter the EEG epochs, the more difficult the detection of a BCI command. Second, BCI commands are often misclassified while the subject is not performing any tasks by not focusing on a command (offset condition) because the EEG characteristics of the offset condition are not unique. In order to circumvent these limitations, the hemodynamic fluctuations in the brain during stimulation with steady-state visual evoked potentials (SSVEP) were measured using near infrared spectroscopy (NIRS) simultaneously with EEG. The offset condition is distinguished from the onset condition (focusing on a command) with extracted NIRS features through the design of low-pass filter. BCI command estimates were based on EEG SSVEP response. Simultaneous evaluation of EEG and NIRS was shown to improve the SSVEP classification, notably including the offset condition as an independent class, using a novel offset condition estimation approach. For 13 subjects, wrong classification for 9 classes with inclusion of offset condition were decreased. This proposed bimodal approach including the offset condition detection may render current BCI systems more reliable.