A neural network will be used to identify stray signals in ionograms. The signals to be identified come from high- frequency (HF, 2--30 MHz) transmitters that emit during the collection of the ionosonde data for the ionogram. Unfortunately, the ionosonde accepts the HF signals as valid and records them as data. The developed neural network correctly identifies 85% of the stray HF signals. ¿ American Geophysical Union 1994