Meander line patterned infrared detectors, with values of resistance R onset at the onset temperature T c onset of 3-5 k⍀, were fabricated from YBa 2 Cu 3 O 7Ϫx superconducting material on MgO and SrTiO 3 crystalline substrates. Noise voltages from the samples were measured versus bias current, radiation modulation frequency, and temperature, in both the normal and superconducting states. Four major types of voltage noise were identified according to where they occurred in temperature relative to T c onset and the zero resistance temperature T c zero , and their dependence on frequency and bias current. They were also associated with the granularity of the superconducting film, which is related to the substrate material used. From these observations a specific cause for each type of noise is suggested. The results are as follows. ͑i͒ In the normal state with temperature T Ͼ T c onset , noise with a magnitude that is consistent with thermal ͑Johnson͒ noise is seen, but it depends linearly on bias current above a threshold value, at low frequencies. The suggested noise source is conductivity fluctuations due to Cooper pairs. ͑ii͒ Noise was found to occur below T c zero in granular films. With increasing bias current its magnitude increases, and it shifts to a lower temperature range; however, the noise magnitude becomes constant as the current goes to zero. It is weakly dependent on frequency above 400 Hz. Suggested cause is voltage fluctuations in superconductor-normal-superconductor junctions at grain boundaries. ͑iii͒ This noise also occurs below T c zero with peaks at various temperatures. With increasing bias current the peaked noise spreads to lower temperatures, but the noise goes to zero as the bias current goes to zero. Its suggested cause is magnetic flux tube motion. ͑iv͒ This noise occurs between T c onset and T c zero and is present in all samples, but lowest on samples prepared on SrTiO 3 substrates. Its suggested cause is fluctuations in the volume fraction of the superconducting phase along the current path. While the measured detectivity D* of our samples at a wavelength of 20 m was only 10 6 cm Hz 1/2 /W, engineering changes can be expected to raise the value to above 10 10 cm Hz 1/2 /W.