One of the major problems concerning latent fingerprint matching in forensic applications is the poor quality of fingerprint data. Therefore, local quality assessment of fingerprint images is necessary to ensure high biometric performance in this application field. While big progress has been made in matching of fingerprints by using local minutiae descriptors invariant to rotation and translation, in particular minutia cylinder-code (MCC), automatic latent fingerprint matching continues to be a challenge. The goal of our research is to develop a matching algorithm that uses minutiae information encoded by MCC with integrated local quality measures. In this paper, firstly, we associate a new local quality measure, called cylinder quality, to each MCC descriptor by combining the qualities of individual minutiae involved. Then, we propose a method for incorporating the cylinder qualities in latent fingerprint matching through a quality-based relaxation procedure in order to cope with challenges originating from poor-quality regions. Experimental results on NIST SD27 show that integrating the cylinder qualities through the proposed method improves the identification performance for latent fingerprints of ugly quality.