The reaction 16O(n, γ) 17O acts as a neutron poison in the weak slow neutron capture process (s process) by reducing the number of available neutrons in the stellar burning environment. The captured neutrons can be re-emitted into the stellar environment via the reaction 17O(α, n) 20Ne, weakening the poisoning effect of 16O. This branch competes with the reaction 17O(α, γ) 21Ne. Therefore in order to determine the strength of 16O as a neutron poison one needs to know the ratio of the two stellar reaction rates 17O(α,γ) 21Ne 17O(α,n) 20Ne.  As there is no published data on 17O(α, γ) 21Ne and only limited information is available on the 17O(α, n) 20Ne reaction both reactions have been measured. The total cross section of the (α, n) reaction was measured using a high efficiency 4π neutron detector. To improve the accuracy of the results the (α, n1) channel has been investigated separately over the same energy range by detecting its characteristic gamma-rays with a germanium detector.  Besides a possible role in the weak s process 18O can be a strong source of beam-induced background in the measurement of (α, n) reactions. Even a very small contamination of the target material with 18O can lead to spurious signals in both the 17O(α, γ) 21Ne and the 17O(α, n) 20Ne measurements.TheAndreas Christian Best reactions 18O(α, n) 21Ne and 18O(α, n1) 21Ne were measured from the threshold up, covering the same energy range as the 17O measurements.  In this work several resonances in 17O(α, γ) 21Ne have been found and their parameters have been determined. The uncertainty in both the 17O(α, n) 20Ne and the 18O(α, n) 21Ne reaction rates has been greatly reduced. The astrophysical implications of the new experimental results are discussed.