This study examines attentional biases in the presence of angry, happy and neutral faces using a modified eye tracking version of the dot probe task (DPT). Participants were 111 young children between 9 and 48 months. Children passively viewed an affective attention bias task that consisted of a face pairing (neutral paired with either neutral, angry or happy) for 500 ms that was followed by a 1,500-ms asterisk probe on 1 side of the screen. Congruent trials were trials in which the probe appeared on the same side of the screen as the emotional face and incongruent trials were trials in which the probe appeared on the opposite side of the emotional face. The latency to fixate on the probe, rather than the traditional task's button press latency, was measured for both types of trials and a bias score was calculated by subtracting the latency to the probe on congruent trials from that on incongruent trials. The results of the current study indicate positive internal reliability of this modified version of the DPT as well as the presence of a bias toward both angry and happy faces during the first 4 years of life. The successful use of the modified version of the DPT for use on the eye tracker presents a promising methodological tool for research on early attentional behavior and provides a tool for comprehensive longitudinal studies of identified risk factors for anxiety.