We have observed a number of strong echoes with the European incoherent-scatter (EISCAT) UHF (930-MHz) radar at angles 83.5° and 78.6° with the geomagnetic field and at about 100-km altitude north in the auroral zone. The echoes are short-lived and occur in single 2- or 10-s data dumps. They are offset by 125–130 kHz with respect to the transmitted frequency. In most cases the offset compares well with the frequency of gyro lines in the incoherent-scatter spectrum, as given by the standard linear dispersion relation. But sometimes the measured offsets deviate significantly from the model calculations, and the interpretation in terms of gyro lines becomes questionable. The discrepancy could possibly be explained by local deviations in the magnetic field from the model (IGRF 1987), which are generated by incoming particle beams. A more serious problem with the gyro-line theory is how the line can be excited at altitudes where the collisional damping is substantial. The high intensity and short lifetime of the signal point to a fast-growing plasma instability as the likely excitation mechanism, if the gyro-line interpretation is correct. The cause of the instability could be the same particle beams as those causing the disturbances in the magnetic field. Alternatively, the observations may be interpreted as meteor head echoes. The large Doppler shifts, the short lifetimes and the altitudes of the signals support this explanation. The main difficulty is that the distribution of measured offsets appears to be different in magnetically active conditions and in less active conditions. Also, the occurrence of echoes does not seem to follow the expected changes in meteor density. More observations in different conditions are needed to decide between the two interpretations. As it is, we are inclined to believe in the meteor head echo theory, the objections to the gyro-line theory being more fundamental.