Global solar radiation data obtained from actinographs of the Turkish State Meterological Service were compared with data obtained from pyrheliometers that were established recently to determine if the actinograph data were usable in practice. It has been found that the observed actinograph data have a rather high error rate with 14.7% annual and 42.1% monthly averages. It has, however, not been possible to smooth these errors as they are absolutely random. Thus, a well known quadratic model was used to produce available radiation data. A quadratic relationship between solar insolation and duration of solar radiation data has been investigated in order to estimate monthly average global irradiance for Ankara, Antalya, Samsun, Konya, Urfa and Izmir. The data from August 1993 to July 1995 have been used in the quadratic model. But, observations of Izmir differ from other stations covered for about four years. Solar insolation data used in the quadratic model were obtained from the pyrheliometer with model CR10. Duration of bright sunshine data were obtained from a Siap, Frans or Muller sunshine recorder with 60° global lens. A general quadratic formula was found that represents the whole of Turkey. The estimated monthly average global solar radiation data, then, were produced from this quadratic formula. Comparison of the estimated and measured values showed that the quadratic model was able to estimate global radiation with about a 4% annual relative error and the estimated data seemed to be more reliable than the data measured by actinograph.