The El Ni¿o-Southern Oscillation (ENSO) is not an extrinsic forcing mechanism on the atmosphere, so its impact thereon cannot be cleanly removed. Nevertheless, it is of interest to estimate what atmospheric temperatures would have been if ENSO had not been operating. This is done here by using a 63-station global radiosonde network to find the relation, for the period 1958--1998, between sea-surface temperature (SST) in the El Ni¿o region of the eastern equatorial Pacific (Nino3 SST) and tropospheric (850--300 mbar) temperature. Seasonal lag correlations for the 41-year interval show maximum (and significant) correlations of 0.67, 0.57, and 0.34 between Nino3 SST and the 850 to 300-mbar temperature in tropics, globe, and north temperate zone, respectively, two seasons later. Accordingly, the annual 850 to 300-mbar temperatures for those regions are adjusted (effect of Nino3 SST on them minimized) by means of the linear least squares relation between the annual values of 850 to 300-mbar temperature and annual values of Nino3 SST shifted two seasons earlier, omitting the years 1964, 1983, and 1992 following Agung, El Chichon, and Pinatubo eruptions, as well as the anomalously warm year of 1998. In 1998 the global 850 to 300-mbar temperature is a record 0.7 K above the 1961--1990 average but, after adjustment in the above manner for the strong 1997--1998 El Ni¿o, only 0.2 K above this average and the twelfth warmest year of record. During 1958--1998 the global 850 to 300-mbar layer warms by 0.10 K/decade, but after adjustment for ENSO events the warming is 0.07 K/decade, so that about one third of the global tropospheric warming is related to ENSO. The adjustment provides a different perspective on the significance of anomalously warm years, as well as warming trends, to the issue of climate change.