An artificial neural network (ANN) method is first used for deriving long-term trends of the F2-layer critical frequency (foF2) at 19 ionospheric stations in the Asia/Pacific sector. It is found that the ANN method can eliminate the geomagnetic activity effect on foF2 more effectively than usual regression methods. Of the selected 19 stations, there are significant long-term trends corresponding to a confidence level ≥90% at 14 stations and 12 of these stations present negative trends. An average trend of -0.05% per year in the selected area can be obtained if the 12 stations with significant negative long-term trends be considered. No pronounced diurnal and latitudinal effects in trends and no uniform pattern of seasonal variation in most stations are detected. The long-term trends for low latitude and equatorial stations differ from other stations suggest that some special dynamical processes may take effects in the equatorial anomaly region. Many factors which can influence ionosphere, such as the greenhouse effect, solar and geomagnetic activity, and neutral background gas, might contribute to the trend.