diff --git a/src/descendPeak.cpp b/src/descendPeak.cpp
new file mode 100644
index 00000000..cf8a1f59
--- /dev/null
+++ b/src/descendPeak.cpp
@@ -0,0 +1,139 @@
+// Pascal Maas - p.maas@lacdr.leidenuniv.nl
+
+#include <Rcpp.h>
+using namespace Rcpp;
+using namespace std;
+
+double weightedMean(NumericVector mzs, NumericVector ints) {
+    int n = ints.length();
+    double total = 0;
+    double intensity = 0;
+    for (int i = 0; i < n; i++) {
+        total += mzs[i] * ints[i];
+        intensity += ints[i];
+    }
+    return total / intensity;
+}
+
+double descend(int centroid, NumericVector intensities,
+               NumericVector masses, double signalRatio, int maxK = 30) {
+
+
+    // Define a min and max index for the region where the region is not negative
+    // and does not exceed the size of the intensities vector, since we only
+    // have to consider centroid - maxK until centroid + maxK
+
+    // Ensures the index is at least 0
+    int minId = std::max(centroid - maxK, 0);
+
+    // Ensures the index is at most the size of the intensities vector - 1
+    int maxId = std::min(centroid + maxK, (int) intensities.size() - 1);
+
+    // Get indices of this region
+    Rcpp::Range idx = Rcpp::Range(minId, maxId);
+
+    // Subset intensities and masses with the defined region
+    NumericVector ints = intensities[idx];
+    NumericVector mass = masses[idx];
+
+
+    // Find the new location of the centroid, as the centroid can be shifted by
+    // subsetting the region. By substracting the minimum value, the original
+    // centroid will now refer to the new centroid location.
+    centroid = centroid - minId;
+    double centroidValue = ints[centroid];
+
+    // Calculate the allowed limits within the region
+    // Update the left and right limits with new idx, as long as the values are
+    // within the thresholds of percentage and are not rising.
+    int leftLimit = centroid;
+
+    // Start looping from centroid to the left (negative)
+    for (int i = centroid - 1; i >= 0; i--) {
+        // Get current and previous values
+        double previous = ints[i + 1];
+        double current = ints[i];
+
+        // Good signal should be above the centroid / signal ratio
+        bool goodSignal =  current / centroidValue > signalRatio;
+
+        // Compare current to previous value to check for a rising signal
+        bool risingSignal = current >= previous;
+
+        // If either the signal is too low or the current signal is rising,
+        // Break the loop since we've reached the left limit
+        if (!goodSignal || risingSignal) {
+            break;
+        }
+        // All good, so update the leftLimit to the current index
+        leftLimit = i;
+    }
+
+    // Repeat this process for the right side of the centroid
+    int rightLimit = centroid;
+    int N = ints.size();
+
+    // Start loop from centroid to right (positive)
+    for (int i = centroid + 1; i < N; i++) {
+        // Get current and previous values
+        double current = ints[i];
+        double previous = ints[i - 1];
+
+        // Good signal should be above the centroid / signal ratio
+        bool goodSignal =  current / centroidValue > signalRatio;
+
+        // Compare current to previous value to check for a rising signal
+        bool risingSignal = current >= previous;
+
+        // If either the signal is too low or the current signal is rising,
+        // Break the loop since we've reached the right limit
+        if (!goodSignal || risingSignal) {
+            break;
+        }
+
+        // All good, so update the rightLimit to the current index
+        rightLimit = i;
+    }
+
+    // Determine the region between the left and right limits
+    // to obtain the region of interest
+    Range region = Rcpp::Range(leftLimit, rightLimit);
+
+    // calculate the weighted mean of the region and return the value
+    return weightedMean(mass[region], ints[region]);
+}
+
+//' @name descendPeaksC
+//' @title descendPeak algorithm from MSnbase for centroid refinement
+//' @description This is the C++ version of `descendPeak` of MSnbase. It
+//' calculates the weighted mean of the centroid region by checking for the
+//' signal ratio and rising signals.
+//' @details DescendPeaks is a centroid-refinement algorithm that descends from
+//' the centroid until the signal rises again. It also considers the intensity
+//' of neighbouring signals to be at least a percentage of the centroid
+//' intensity.
+//' @param centroids Position of the centroids (local maximum) as C-index
+//' (R-index - 1).
+//' @param intensities Vector of intensities in a scan, preferably smoothed
+//' @param masses Vector of masses in a scan
+//' @param signalPercentage Percentage between 0-100 defining the minimum signal
+//' for a peak to be considered for the weighted intensity region.
+//' @return Weighted masses for the given centroids
+//' @export
+// [[Rcpp::export]]
+NumericVector descendPeaksC(NumericVector centroids, NumericVector intensities,
+                           NumericVector masses, double signalPercentage, int maxK = 30) {
+
+    // Prepare variables
+    int n = centroids.length();
+    double signalRatio = signalPercentage / 100;
+    NumericVector weightedMZs(n);
+
+    // Loop through all centroid positions (should be C-indexed, not R-indexed!)
+    for (int i = 0; i < n; i++) {
+        // Calculate the weighted mass
+        weightedMZs[i] = descend(centroids[i], intensities,
+                                 masses, signalRatio, maxK);
+    }
+    return weightedMZs;
+}