Predictions of volcanic eruptions, generally speaking, can not be achieved by observations of a single kind. They depend on synthetic consideration of the observational results of various kinds. Geodetic, geomagnetic, geoelectric and geothermic observations, that is, miscellaneous methods besides seismometric and geochemical ones, should be indispensable to achievement of the predictions. Geodetic observations consist of precise levellings, distance measurements and continuous observations of tilt. The intermittent surveys are useful for long-term prediction while the continuous observations are necessary for short-term prediction. Gravity changes at volcanic areas may be caused by density changes of subsurface material and vertical displacements of the earth-surface. In many volcanic areas, gravity measurements have been repeated several times with accuracies of ±20μgal, but not favoured by any sufficiently large changes. Geomagnetic observations would be carried out by proton precession magnetometers with accuracies of ±1γ. Their results have raised new difficulties in eliminating the disturbances of external origin and the local characteristics of secular changes. Geoelectric methods aim at detection of hot material by measuring subsurface conductivity. They are worthy of practice near the craters which are accessible with an apparatus. Geothermic methods are recently conccntrated on infrared imageries, especially air-borne ones. These three years, the air-borne infrared surveys have been carried out over about 20 volcanoes in Japan in order to prepare fundamental data for prediction of volcanic eruptions.