We present spot models for nine years of continuous VIC photometry of UZ Lib from 1993-2001. The relatively stable double-humped light curve shape suggests extreme phase coherence. From the spot-modelling analysis, we found that the major spots or spot groups are always located on the hemisphere facing the secondary star and exactly in the opposite hemisphere anti-facing the secondary. Several single-humped light curves and our suggested binary scenario rule out a pure ellipsoidal variability as the cause of the double-humped light curve shape. We try to explain this preferred spot pattern with a magnetic-field structure that connects the two components, as suggested earlier for RS CVn stars in general. A possible 4.8 years spot cycle is found from the long-term brightness variations but needs confirmation. We rediscuss the basic astrophysical data of UZ Lib. The Hipparcos parallax is likely wrong, a possible reason could be that UZ Lib is e.g., a triple system.