Résumé : Despite decades of creatinine measurement in biological fluids using a large variety of analytical methods, an accurate determination of this compound remains challenging. Especially with the novel trend to assess biomarkers on large sample sets preserved in biobanks, a simple and fast method that could cope with both a high sample throughput and a low volume of sample is still of interest. In answer to these challenges, a fast and accurate ultra-high performance liquid chromatography–tandem mass spectrometry (UHPLC–MS/MS) method was developed to measure creatinine in small volumes of human urine. In this method, urine samples are simply diluted with a basic mobile phase and injected directly under positive electrospray ionization (ESI) conditions, without further purification steps. The combination of an important diluting factor (104 times) due to the use of a very sensitive triple quadrupole mass spectrometer (XEVO TQ) and the addition of creatinine-d3 as internal standard completely eliminates matrix effects coming from the urine. The method was validated in-house in 2012 according to the EMA guideline on bioanalytical method validation using Certified Reference samples from the German External Quality Assessment Scheme (G-Equas) proficiency test. All obtained results for accuracy and recovery are within the authorized tolerance ranges defined by G-Equas. The method is linear between 0 and 5 g/L, with LOD and LOQ of 5 × 10−3 g/L and 10−2 g/L, respectively. The repeatability (CVr = 1.03–2.07%) and intra-laboratory reproducibility (CVRW = 1.97–2.40%) satisfy the EMA 2012 guideline. The validated method was firstly applied to perform the German G-Equas proficiency test rounds 51 and 53, in 2013 and 2014, respectively. The obtained results were again all within the accepted tolerance ranges and very close to the reference values defined by the organizers of the proficiency test scheme, demonstrating an excellent accuracy of the developed method. The method was finally applied to measure the creatinine concentration in 210 urine samples, coming from 190 patients with a chronic kidney disease (CKD) and 20 healthy subjects. The obtained creatinine concentrations (ranging from 0.12 g/L up to 3.84 g/L) were compared, by means of a Passing Bablok regression, with the creatinine contents obtained for the same samples measured using a traditional compensated Jaffé method. The UHPLC–MS/MS method described in this paper can be used to normalize the concentration of biomarkers in urine for the extent of dilution.