In recent years, eye tracking has become a more popular input modality and is utilized in both assistive applications and immersive computer games. Related work within the subject indicate that the implementation of gaze-awareness can increase efficiency in some applications, but also that results vary reatly between different types of applications and types of technical implementations. This paper aims to investigate how well eye tracking performs in a rhythm game when compared to input from a tradition pointing device (mouse) at different stages of perceived speed and limited reaction time. For the study, a gaze-based video game called Gaze Hero has been developed, in which the player must hit notes that timely appear in a one-dimensional array, similar to gameplay in the Guitar Hero game series. From the evaluation of the game, in which 16 participants played on 3 different levels of perceived game speed, no evidence was found to support that using gaze as input modality results in better performances in a game where timely precision and continuous saccadic motions are required with high accuracy. Contrarily, participants performed better when using mouse as input in fast perceived situations where objects move fast and require quick reaction times. However, in situations where the player has a greater time to react and plan upcoming saccadic motions, no statistical significant difference was found in the performance between gaze and mouse. This suggests that gaze input may be a sufficient substitution for interactions in which the user has at least 1.14 seconds to react and plan gaze movement.