Climate change and rapid urbanization have significantly altered the hydrological conditions of cities, making urban flooding and runoff a major environmental hazard. Urban flooding is a multi-dimensional problem with economic, social, and environmental consequences. In response, numerous studies have been conducted in Iran, focusing on hydrological/hydraulic modeling (e.g., SWMM, HEC-HMS, HEC-RAS) or GIS-based multi-criteria decision-making (MCDM) for flood hazard zoning. However, most research concentrates on physical dimensions (depth, extent, drainage performance) while neglecting social, economic, and physical vulnerability components. Furthermore, the diversity of models without a coherent analytical framework has led to fragmented knowledge. This systematic review aims to answer: What approaches, models, and analytical frameworks have been used in Iranian urban flood management studies? What input data and validation methods are employed? What are the strengths and limitations of these approaches? This mixed-method (quantitative-qualitative) systematic review follows PRISMA guidelines. A comprehensive search of Persian-language databases (Civilica, IranDoc, Google Scholar) using keywords related to urban flood, runoff, risk management, vulnerability, and GIS was conducted in January 2026 (Dey 1404). The initial search identified 112 records. After removing duplicates (n=10), screening for relevance (n=12), and excluding inaccessible full texts (n=5), 85 records were assessed. Following a full-text review against inclusion criteria (clear thematic relevance, use of hydrological/topographic/land-use data, link to urban planning/flood management), 18 records were excluded, leaving 67 studies for final synthesis. Data were extracted using a coding protocol in Excel. Analysis combined descriptive quantitative statistics (temporal distribution, spatial scales, models) with qualitative thematic analysis to identify conceptual patterns and develop an integrated framework. The 67 sources comprised 50 research articles (74.6%), 10 conference papers (14.9%), 6 master's theses (9.0%), and 1 doctoral dissertation (1.5%). Temporal distribution shows gradual growth from 2007-2008, with a significant increase after 2017. The highest publication year was 2018 (13.4%), followed by 2024 (11.9%). Subject areas were categorized into six themes: quantitative runoff simulation, spatial flood hazard/vulnerability assessment using GIS-MCDM, green infrastructure/Low Impact Development (LID) performance, social/institutional dimensions (limited), remote sensing applications, and data-driven machine learning models (emerging). The city scale was most frequent (32 studies), followed by sub-city (e.g., districts) and beyond-city (e.g., watersheds). Geographically, Tehran dominated, followed by Mashhad, Tabriz, and others, indicating uneven spatial distribution. Five model categories were identified. (1) Hydrological models (SWMM, HEC-HMS, SCS-CN) were widely used for rainfall-runoff simulation. Their strength is quantitative estimation of runoff and peak flow, but accuracy depends on input data quality and calibration. (2) Hydraulic models (HEC-RAS 1D/2D, StormCAD, SewerGEMS) enabled analysis of flow behavior, water depth, and inundation extent, making them essential for hazard zoning. (3) Ecosystem service models (InVEST, regulatory service frameworks) were less common, offering an interdisciplinary view linking floods with ecological structure and resilience, though with lower spatial-physical precision. (4) MCDM models (AHP, ANP, TOPSIS, COPRAS, PROMETHEE II, fuzzy logic combined with GIS) were heavily used for hazard zoning, vulnerability assessment, and site selection. Their strength lies in integrating quantitative and qualitative criteria, but results depend on expert judgment. (5) Machine learning models (Maxent, PLS, Ridge Regression, LSTM) are emerging, capable of representing non-linear patterns, but suffer from lower interpretability and generalization challenges with limited data. Key inputs included topographic data (DEM for slope, flow direction), rainfall/hydrological data (often limited sub-hourly data), drainage network data, observational flood data (rare but crucial for calibration), climate and scenario data (very limited climate scenarios; mostly land-use or LID management scenarios), remote sensing data (Landsat, Sentinel), and integrated datasets for MCDM. Weak validation and lack of observational data were recurring issues. Only 23.9% of studies had full or adequate validation. 43.3% had limited validation, and 31.3% lacked explicit validation. Validated studies used statistical methods (NSE, RMSE, R²), consistency checks, or spatial mapping against historical events. The absence of validation is a critical flaw, undermining reliability. Most studies (41) made no official reference to ecosystem services. Sixteen implicitly referred to ecosystem functions, 8 made explicit but non-framework references, and only 2 used a formal ecosystem service framework. Attention to Nature-based Solutions (NbS) was similarly low, with most studies not simulating NbS scenarios. Regulating services (flood risk reduction, runoff regulation) dominated, while multi-service approaches and socio-economic integration were rare. Most studies remained purely technical (31), with 13 adopting multi-dimensional approaches. The findings indicate a dominant engineering paradigm with limited integration of ecological and socio-economic dimensions. The literature follows a dominant engineering-technical paradigm, treating floods as a primarily hydrological problem. This conceptual narrowness neglects social, institutional, and ecological complexities. There is a disconnect between physically accurate models (weak in socio-economic integration) and MCDM models (subjective but integrative). Poor validation and scarce observational data create a gap between research and operational application. Compared to international trends (integrated flood management, urban resilience, NbS), Iranian research has not fully transitioned. The concentration of studies in major cities (Tehran) reflects data and infrastructure disparities. Limited access to high-temporal-resolution rainfall data and historical flood records hinders validation. The past two decades have seen growth in Iranian urban flood research, but the field suffers from conceptual and methodological heterogeneity. The dominant engineering-technical paradigm needs to shift toward integrated, multi-dimensional frameworks (social, economic, institutional, ecological). While GIS-MCDM methods are widespread, they often lack validation and transparency. Emerging machine learning and NbS approaches signal a paradigm shift, but it remains immature. Key policy implications include: moving toward integrated risk management, strengthening data infrastructure (high-resolution rainfall, observational flood data), developing model validation standards, and promoting NbS/green infrastructure. Research limitations include the exclusive focus on Persian-language databases, heterogeneity in study quality, and potential interpretive bias. Future research should focus on: integrating physical models with socio-economic analysis; conducting empirically validated studies; applying ecosystem service and NbS frameworks in scenario-based studies; hybridizing machine learning with physical models; and expanding geographic coverage to understudied cities. The persistence of urban flooding in Iran is not due to a lack of studies but to the gap between knowledge production, methodological quality, and urban decision-making processes.