The physical, chemical, and mechanical properties of soil are crucial for implementing appropriate land management strategies. Among these, phosphorus is one of the most essential nutritional elements for plant growth. The study aimed to (i) evaluate the potential of visible–near-infrared (Vis–NIR) spectroscopy for predicting soil available phosphorus and (ii) investigate the effects of different spectral preprocessing techniques on prediction accuracy. A total of 150 samples were collected from tea farms in Lahijan County and from agricultural fields around the Tehran Refinery (Ismailabad) and the refinery south of Tehran (Ray City). Reflectance visible-near infrared (Vis-NIR) spectrometry for measurement and prediction of soil phosphorus content was studied. To this end, spectra of air-dried soil samples were acquired using a FieldSpec® 3 ASD spectroradiometer equipped with a contact probe, operating in interactance measurement mode over the spectral range of 350–2500 nm. The spectra were preprocessed using Savitzky–Golay smoothing, first (D1) and second (D2) derivatives, Multiplicative Scatter Correction (MSC), Moving Average (MA), and Standard Normal Variate (SNV) transformation. The average of five replicate measurements per sample was calculated and used for subsequent analysis. The results demonstrated that Vis–NIR spectroscopy is capable of predicting soil phosphorus content with satisfactory accuracy. The PLS model combined with SG + MSC preprocessing yielded the best predictive performance (R²c = 0.94, RMSEC = 13.04, R²p = 0.91, RMSEP = 16.00, SDR = 3.28). Nevertheless, other preprocessing combinations also showed promising potential for phosphorus prediction, indicating the flexibility of the PLS model with different spectral transformations.