How to select DataFrame rows and columns with loc and iloc in pandas

Selecting rows and columns with loc and iloc in pandas extracts a precise subset from a DataFrame before analysis, cleanup, joining, or export. The two indexers make the selection rule explicit, so label-based row IDs and positional table coordinates do not get mixed.

Use loc when row labels, column labels, or a boolean condition identify the subset. Label slices include both the start and stop labels, and a boolean Series aligns by index before masking.

Use iloc when integer positions identify the subset independently of labels. Position slices use normal Python slice boundaries, so the stop position is excluded; check the resulting shape, labels, and key values before feeding the subset into later code.

1. Save a short loc and iloc selection script.

   select_loc_iloc.py

   import pandas as pd
    
   orders = pd.DataFrame(
       {
           "order_id": ["A101", "A102", "A103", "A104", "A105"],
           "customer": ["Ada", "Lin", "Maya", "Omar", "Nia"],
           "region": ["EMEA", "APAC", "AMER", "EMEA", "APAC"],
           "qty": [3, 12, 7, 2, 15],
           "total_usd": [150.0, 240.0, 875.0, 95.0, 360.0],
           "status": ["paid", "paid", "open", "paid", "open"],
       }
   ).set_index("order_id")
    
   print(f"pandas {pd.__version__}")
   print()
    
   print("BASE")
   print(orders.to_string())
   print()
    
   label_rows = ["A101", "A104"]
   label_columns = ["customer", "total_usd"]
   label_list = orders.loc[label_rows, label_columns]
    
   print("LOC_LABEL_LIST")
   print(label_list.to_string())
   print()
    
   label_slice = orders.loc["A102":"A104", ["region", "qty"]]
    
   print("LOC_LABEL_SLICE")
   print(label_slice.to_string())
   print()
    
   position_rows = [0, 3]
   position_columns = [0, 3]
   position_list = orders.iloc[position_rows, position_columns]
    
   print("ILOC_POSITION_LIST")
   print(position_list.to_string())
   print()
    
   position_slice = orders.iloc[1:4, 1:3]
    
   print("ILOC_POSITION_SLICE")
   print(position_slice.to_string())
   print()
    
   print("VERIFY")
   print(f"label rows: {label_list.index.tolist()}")
   print(f"position shape: {position_list.shape}")
   print(f"label slice includes A104: {'A104' in label_slice.index}")
   print(f"iloc row labels: {position_slice.index.tolist()}")
   print(f"source shape unchanged: {orders.shape}")
   print(f"A104 total by loc: {orders.loc['A104', 'total_usd']}")
   print(f"A104 total by iloc: {orders.iloc[3, 3]}")

   The order_id index makes row labels distinct from row positions. In a working script, replace the sample orders object with the DataFrame already loaded.

2. Run the script and compare the label-based and position-based selections.

   $ python3 select_loc_iloc.py
   pandas 3.0.3
   
   BASE
            customer region  qty  total_usd status
   order_id                                       
   A101          Ada   EMEA    3      150.0   paid
   A102          Lin   APAC   12      240.0   paid
   A103         Maya   AMER    7      875.0   open
   A104         Omar   EMEA    2       95.0   paid
   A105          Nia   APAC   15      360.0   open
   
   LOC_LABEL_LIST
            customer  total_usd
   order_id                    
   A101          Ada      150.0
   A104         Omar       95.0
   
   LOC_LABEL_SLICE
            region  qty
   order_id            
   A102       APAC   12
   A103       AMER    7
   A104       EMEA    2
   
   ILOC_POSITION_LIST
            customer  total_usd
   order_id                    
   A101          Ada      150.0
   A104         Omar       95.0
   
   ILOC_POSITION_SLICE
            region  qty
   order_id            
   A102       APAC   12
   A103       AMER    7
   A104       EMEA    2
   
   VERIFY
   label rows: ['A101', 'A104']
   position shape: (2, 2)
   label slice includes A104: True
   iloc row labels: ['A102', 'A103', 'A104']
   source shape unchanged: (5, 5)
   A104 total by loc: 95.0
   A104 total by iloc: 95.0

   LOC_LABEL_SLICE includes A104 because loc label slices include the stop label. ILOC_POSITION_SLICE stops before row position 4 but still carries the original row labels in the result.