This paper addresses a standard paleoclimatological approach to reconstructing temperatures in historical climatology. Weighted monthly temperature indices on a scale from - 3 to + 3, derived from documentary evidence for the period 1718-1850 in the Czech Lands, were used to create a series of seasonal (DJF, MAM, JJA, SON) and annual temperature indices as simple sums of indices from corresponding months. The period 1771-1850, when such indices overlap with instrumental measurements taken at Prague-Klementinum, was used to calibrate and verify relations between indices and the temperatures measured, using linear regression models (LRM) evaluated by various statistical characteristics. The most accurate LRM performance was obtained for DJF, MAM and annual temperatures, while JJA and SON temperature results proved slightly less reliable. Reconstructed data for 1718-1770 were combined with measured data to create series of seasonal and annual temperatures for the Prague-Klementinum station covering the period 1718-2007. Various problems in temperature reconstruction are further discussed: the completeness and quality of the index series, the expression of extreme temperatures, and the selection of a reference period for calibration. The suggested standard paleoclimatological approach lends more objectivity to both verification of reconstructed temperatures and the expression of uncertainties in reconstruction.