The traveltimes of interplanetary (IP) shocks at 1 AU associated with coronal mass ejections (CMEs) can be predicted by the empirical shock arrival (ESA) model of Gopalswamy et al. [2004] based on a constant IP acceleration. We evaluate the ESA model using 91 IP shocks identified from sudden commencement (SC)/sudden impulse (SI) on the Earth and by examining the solar wind data from the ACE and WIND satellites during the period of 1997 to 2002. Out of 91 CME-IP shock pairs, 55 events (∼60%) were predicted within ±12 hours from the ESA model. The ESA model predicted ∼59% (43 out of 73) of the events during solar maximum (1999–2002) and ∼67% (12 out of 18) of the events during solar minimum (1997–1998) within ±12 hours from the predicted curve. Comparing the predicted (T _mod) and observed (T _obs) shock arrival times during solar maximum, we find that the deviations (ΔT = T _obs − T _mod) of shock arrival times from the ESA model strongly correlate with the CME initial speeds (V _CME) (linear correlation, r = 0.77). Such a strong correlation indicates that the constant IP acceleration in the ESA model is not reasonably well applied for all V _CME. From the linear regression analysis, we obtain a linear fit to the relationship (r = −0.62) between IP shock traveltime T (in hours) and V _CME (in kilometer per second) during the solar maximum, which can be expressed as T = 76.86 − 0.02V _CME. In addition, we find that the IP shocks associated with the fast CMEs corresponding to strong SC/SI events have short traveltimes compared with other fast CMEs and that there is a negative correlation between the SC/SI strength and the IP shock traveltime. We suggest that this negative correlation is due to not only the V _CME but also the CME mass/density and discuss the influence of the mass/density of CME on the arrival time of IP shock at 1 AU.