This dataset contains element to calcium ratios (Mg/Ca, Sr/Ca, Mn/Ca, Fe/Ca, Ba/Ca and Al/Ca in mmol/mol) for benthic foraminifera obtained from gravity core GeoB9506-1 (Mulitza et al., 2005). Data is presented for 296 samples, consisting of Uvigerina peregrina (222 samples), Uvigerina mediterranea (7 samples) and Melonis barleeanus (67 samples). 15 duplicate samples (indicated in column 15 "Duplicates") are included. Sediment samples came from both previously washed, and fresh; mud Fresh samples were washed using deionized water through a 63 µm sieve, dried at 45°C for less than 24 hours, and dry sieved. Benthic foraminifera were picked from the >150 µm fraction. Selected samples were weighed on a microbalance, and to avoid calibration problems, samples with weights over 400 µm were split into smaller sample sizes ("Splitted Sample" in the dataset), and the estimated new weight was calculated using the average specimen weight of each samples and multiplying it by the total specimens picked for analyses ("Estimate test mass used" in the data set). For duplicates, an equal number of specimens was crushed separately for the original and duplicate measurement. If weighed, each portion was assumed to have half the mass of the total sample. In general, between 3 and 11 specimens were used for U. peregrina and between 10 and 16 specimens for M. barleeanus. Before analysis, foraminifera samples were crushed and cleaned with water, methanol and hot buffered hydrogen peroxide with no reductive cleaning step (Johnstone et al., 2016). Samples were measured using an Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES)—Agilent Technologies 700 Series with Cetac ASX-520 autosampler at the following spectral lines: Ca (315.9 nm), Mg (279.6 nm), Sr (421.6), Fe (238.2 nm), and Mn (257.6 nm), Ba (455.4 nm) and Al (167.0 nm) calibrated using linear regressions. Calibration standards consisted of a blank and five multi-element standards of between 20 and 100 ppm Ca. Instrumental precision was monitored by using a commercial multi-element standard solution (SCP, France), and commercial powder sample ECRM752-1. Theoretical value (measured value, σ standard deviation, standard error) in mmol/mol for the SCP standard during these runs were: Mg/Ca, 2.955 (2.959, σ 0.030, 0.0029), Sr/Ca, 1.402 (1.402, 0. σ 016, 0.0015), Mn/Ca 0.3234 (0.3291, σ 0.0019, 0.00019), Fe/Ca, 0.3312 (0.3228, σ 0.0034, 0.00033), Ba/Ca, 0.00832 (0.00840,  0.00034, 0.000033) and Al/Ca 0.678 (0.695, σ 0.018, 0.0018), n=108. Mg/Ca of dissolved, centrifuged, ECRM 752-1 (measured at 3.750 σ 0.015, 0.0014 by Greaves et al. (2005)) was 3.735 σ 0.031, 0.0057. Bottom water temperatures (BWT) were estimated using the calibrations by: (1) Elderfield et al. (2010) (Mg/Ca= 0.81 + 0.074BWT) and Roberts et al. (2016) (Mg/Ca=0.843 + 0.0915T) for Uvigerina species Mg/Ca and (2) Kristjánsdóttir et al. (2007) (Mg/Ca=0.658 * exp [0.137*BWT]) for Melonis barleeanus Mg/Ca. Median ages for the interval between 2.5 and 260 cm core depth were modelled using the Bacon script (Blaauw and Christen, 2011) in RStudio and the calibration curve Marine20 (Heaton et al., 2020), based on the published radiocarbon ages from the same site (Barragán-Montilla and Mulitza, in review).