The analysis of the light curves of 48 B-type stars observed by Kepler is presented. Among these are 15 pulsating stars, all of which show low frequencies characteristic of SPB stars. Seven of these stars also show a few weak, isolated high frequencies and they could be considered as SPB/beta Cep hybrids. In all cases the frequency spectra are quite different from what is seen from ground-based observations. We suggest that this is because most of the low frequencies are modes of high degree which are predicted to be unstable in models of mid-B stars. We find that there are non-pulsating stars within the beta Cep and SPB instability strips. Apart from the pulsating stars, we can identify stars with frequency groupings similar to what is seen in Be stars but which are not Be stars. The origin of the groupings is not clear, but may be related to rotation. We find periodic variations in other stars which we attribute to proximity effects in binary systems or possibly rotational modulation. We find no evidence for pulsating stars between the cool edge of the SPB and the hot edge of the delta Sct instability strips. None of the stars show the broad features which can be attributed to stochastically-excited modes as recently proposed. Among our sample of B stars are two chemically peculiar stars, one of which is a HgMn star showing rotational modulation in the light curve.

We have used Lomb-Scargle periodogram analysis and Monte Carlo significance tests to detect periodicities above the 3-sigma level in the Beta Cephei stars V400 Car, V401 Car, V403 Car and V405 Car. These methods produce six previously unreported periodicities in the expected frequency range of excited pulsations: one in V400 Car, three in V401 Car, one in V403 Car and one in V405 Car. One of these six frequencies is significant above the 4-sigma level. We provide statistical significances for all of the periodicities found in these four stars.

We discuss methods currently in use for determining the significance of peaks in the periodograms of time series. We discuss some general methods for constructing significance tests, false alarm probability functions, and the role played in these by independent random variables and by empirical and theoretical cumulative distribution functions. We also discuss the concept of "independent frequencies" in periodogram analysis. We propose a practical method for estimating the significance of periodogram peaks, applicable to all time series irrespective of the spacing of the data. This method, based on Monte Carlo simulations, produces significance tests that are tailor-made for any given astronomical time series.