Background: Holistic face processing is an important component of both familiar and unfamiliar face processing. Unfamiliar identity recognition is impeded but not eliminated under conditions where the lower half of a face is obstructed. There has been no research done on whether holistic face processing still occurs when an unfamiliar face is partially obstructed. Purpose: The present study sought to address whether participants can accurately match the bottom halves of two different people’s faces with the correct top half. Method: The current experiment used a facial feature matching task in which participants were asked which of two bottom halves appeared most correct when paired with a target top half. Thirty-nine participants were recruited from the Psychology Participant Pool at the University of Regina. A classic onesample t-test and Bayesian one-sample t-test were used to determine whether participants’ accuracy was better than expected by chance. Results: Participants performed significantly better than expected by chance, t(38) = 19.310, p < .001. Bayesian analysis revealed extreme evidence in favour of the hypothesis that participants can accurately match the facial features in the top half of a face with the facial features from the corresponding bottom half of the face, BF10 = 1.15e+38. Implications: This study is a precursor to another study examining whether participants create accurate and consistent expectations of obstructed facial features based on existing available facial features. The present study suggests that visible facial features may be used to guide expectations of obstructed facial features.