We use a continuous 30 day incoherent scatter radar experiment at Millstone Hill in October 2002 to examine day‐to‐day thermospheric variability in exospheric temperature Tex. Solar flux and magnetic activity influences as the main driving factors for day‐to‐day variability are investigated quantitatively. Solar ultraviolet flux levels are based on the TIMED/SEE space weather product, allowing for analysis of ultraviolet flux‐Tex correlation. Tex is most sensitive to solar EUV flux with approximately a 2 day delay at wavelengths of 27–34 nm (including 30.4 nm). In particularly, a 20–60 h time delay occurs in Tex response to EUV flux at 27–34 nm band, with shorter delays in the morning and longer delays in the afternoon and at night. The 1 ∼ 2 day delayed Tex response to solar ultraviolet flux and associated thermospheric solar preconditioning (“memory”) are most significant in the daily mean for the 27–34 nm band, in the diurnal and semidiurnal amplitudes for the soft X‐ray flux at 0.1–7 nm, and in the diurnal amplitude for longer wavelengths. An empirical model driven only by EUV flux at 27–34 nm from 2 days in advance reproduces 90% of the observed variability in the Tex daily mean. With a 2 day time delay, solar X‐ray flux at 0.1–7 nm is correlated positively with Tex diurnal amplitude and negatively with Tex semidiurnal amplitude. Finally, magnetic activity control, as represented by the Dst index, is weaker during the day and stronger at night and is important for the semidiurnal amplitude but not important for the daily mean. Key Points Correlation analysis between exospheric temperature and EUV at different bands A 20–60 h time delay in exospheric temperature response to EUV flux EUV‐only model at 27–34 nm for 2 days in advance produces 90% of DC variability