This dataset contains full raw spectral reflectance data collected at the canopy level of winter wheat (Triticum aestivum L., variety: Bc Opsesija) to assess aphid infestation levels and support machine learning classification models. The data were collected from experimental winter wheat fields (45°44′50.0″N, 15°56′18.8″E) during the 2022 growing season. Study Period: Mid-April to mid-June 2022Location: Bc Institute, Zagreb, CroatiaField Design: 8 plots (10 × 3 m each): 4 plots treated with insecticides (no insect infestation), 4 plots untreated (natural aphid infestation development)Growth Stages Monitored:T1 (Stem elongation–Heading, BBCH 30–55)T2 (Flowering, BBCH 55–59)T3 (Milky grain development, BBCH 69–75)Aphid species recorded: Sitobion avenae, Rhopalosiphum padi, Metopolophium dirhodum, Schizaphis graminumAssessment method: Visual inspection using a magnifying glass and aphid counts recorded per tillerClassification into healthy (below ET) and infested (above ET) categoriesThe economic thresholds (ETs) for aphid infestation were determined based on the plant's growth stage: During T1, the threshold was set at 4 aphids per tiller. At T2, the threshold ranged between 4 and 7 aphids per tiller. During T3, the threshold increased to 8–12 aphids per tiller.Tillers with aphid densities below these thresholds were classified as healthy, whereas those exceeding the thresholds were categorized as aphid-infested.Spectral Data Collection:Instrument: Spectral Evolution® SR-2500 portable spectroradiometerSpectral Range: 350–2500 nmResolution: VIS-NIR (350–1000 nm): 5 nm, SWIR (1000–2500 nm): 22 nmMeasurement Conditions: Taken under sunny, cloud-free, windless conditions. Conducted between 10:00–14:00 for consistent lighting.Calibration: White BaSO₄ reference panel used every 10 scansSampling Approach: Pistol grip fiber holder positioned 50 cm above canopy 25° field of view, covering ~0.2 m² per scanData recorded using Darwin SP V1.5 software.File Structure: Each measurement period (T1, T2, T3) contains two datasets, one for aphid-infested wheat and one for healthy wheat. This raw dataset provides a valuable resource for researchers working on crop stress detection, aphid infestation monitoring, precision agriculture, and spectral data-driven machine learning applications. It can be used for developing vegetation indices, threshold-based detection methods, and spectral classification models.