ABSTRACT Motivated by the alleged detection of a cyclotron resonance scattering feature in the X-ray spectrum of the Be-transient KS 1947+300 with NuSTAR, not observed with other observatories, we have revisited the available observations of the source to clarify the origin of the reported feature. We focus on the analysis of observations obtained with the BeppoSAX observatory during the 2001 outburst, and by NuSTAR in 2013, when the discovery of the line was reported. We find that in both cases, the source broad-band X-ray continuum can be well described with a two-component Comptonization model, which does not require an additional absorption-like feature. In addition, we show that low-significance residuals mimicking the line, which appear using different continuum models, are consistent with known calibration uncertainties. We therefore conclude that the reported cyclotron line detection is largely questionable, and thus the question on the strength of the source magnetic field remains open. We briefly discuss the origin of the observed broad-band continuum in the context of the pulse-phase-resolved analysis conducted for BeppoSAX data and speculate that the soft component may be associated with the emission from polar areas of the pulsar irradiated by the accretion column.