Molodenskii's solution of the geodetic boundary-value problem is approximately derived by means of G1, a kind of terrain correction to gravity anomaly. Although many geodesists have theoretically discussed G1 since Molodenskii's problem was introduced, a very few reports of practical computation of G1 have been presented. In the present paper, the author practically obtains G1 using the gravimetric and terrain data of Tanzawa Mountains. The conclusions obtained here are summarized as follows: (a) In numerical computations of G1, an integration range of 20 km from gravity station can effectively suppress errors under 0.5 mgals In case of a gently-sloping hill, an integration with a range of 10 km works well in accurately calculating G1. (b) The contribution of the inner zone of the integration range to G1 is so large, that detailed maps of topography and gravity anomaly in the inner zone are necessary for data-reading with high accuracy. (c) Gi mirrors short-wavelength free-air gravity or terrain relief. G1 can also be approximated by a quantity proportional to the vertical gradient of gravity (free-air) anomaly. (d) Assuming as a rough estimate, we see that G1 produces errors of about 6 cm at maximum in the height anomaly and about 10 sec in the gravimetric deflection of the vertical on Tanzawa Mountains. The error estimated in the deflection of the vertical is not small as compared with the astrogeodetic deflections observed in the neighborhood of this area.