We have produced, using objective techniques, a long-term series of average Northern Hemisphere temperatures based on monthly mean station data gridded on a 5° latitude by 10° longitude grid. Difficulties in the estimation of this parameter are discussed, deficiencies in the currently available data base and possible effects on the estimated average are described, and monthly mean data are presented. Long-term trends and extremes are identified in the annual and seasonal data. All seasons show similar long-term trends, but there are noticeable differences on time scales of 10 years and less. For example, for winter temperature, the early 20th century warming peaked during the 1940's whereas the maximum in the other seasons was in the previous decade. Both the magnitude of the long-term trends and the year-to-year variability has been greatest in winter. There is evidence that the long-term cooling that characterized the 1940's, 1950's and 1960's has ended. Warming began in the mid to late 1960's ... Abstract We have produced, using objective techniques, a long-term series of average Northern Hemisphere temperatures based on monthly mean station data gridded on a 5° latitude by 10° longitude grid. Difficulties in the estimation of this parameter are discussed, deficiencies in the currently available data base and possible effects on the estimated average are described, and monthly mean data are presented. Long-term trends and extremes are identified in the annual and seasonal data. All seasons show similar long-term trends, but there are noticeable differences on time scales of 10 years and less. For example, for winter temperature, the early 20th century warming peaked during the 1940's whereas the maximum in the other seasons was in the previous decade. Both the magnitude of the long-term trends and the year-to-year variability has been greatest in winter. There is evidence that the long-term cooling that characterized the 1940's, 1950's and 1960's has ended. Warming began in the mid to late 1960's ...