diff --git a/backend/src/services/dataset/service.py b/backend/src/services/dataset/service.py
index 9fba906..a70387c 100644
--- a/backend/src/services/dataset/service.py
+++ b/backend/src/services/dataset/service.py
@@ -14,6 +14,7 @@
     StorageError,
     ValidationError,
 )
+from src.domain.adapters import CSVSchemaDetector
 from src.repo.dataset import DatasetRepo
 from src.schemas.db import Datasets
 from src.services.storage import storage
@@ -271,6 +272,120 @@ async def get_dataset_files(
 
         return files_data
 
+    async def drop_zero_enrollment(
+        self, dataset_id: UUID, user_id: UUID
+    ) -> dict[str, pd.DataFrame]:
+        """
+        Return dataset files with zero-enrollment courses removed.
+
+        Args:
+            dataset_id: Dataset ID
+            user_id: User ID for authorization
+        
+        Returns:
+            Dictionary with filtered courses, enrollments, and rooms dataframes
+        """
+        files = await self.get_dataset_files(dataset_id, user_id)
+
+        courses_df = files["courses"]
+        enrollments_df = files["enrollments"]
+
+        filtered_courses_df, allowed_crns = self._filter_nonzero_enrollment(
+            courses_df
+        )
+
+        # If we couldn't determine CRNs/columns, keep enrollments as-is.
+        filtered_enrollments_df = (
+            self._filter_by_allowed_crns(enrollments_df, allowed_crns)
+            if allowed_crns is not None
+            else enrollments_df
+        )
+
+        return {
+            "courses": filtered_courses_df,
+            "enrollments": filtered_enrollments_df,
+            "rooms": files["rooms"],
+        }
+
+    def _filter_nonzero_enrollment(
+        self, courses_df: pd.DataFrame
+    ) -> tuple[pd.DataFrame, set[str] | None]:
+        """
+        Filter the courses DataFrame to remove rows where Total_Enrollment == 0.
+
+        Returns:
+            (filtered_df, allowed_crns)
+        """
+        try:
+            schema, column_mapping = CSVSchemaDetector.detect_schema_version(
+                courses_df, "courses"
+            )
+        except Exception:
+            # If schema detection fails, don't change behavior.
+            return courses_df.copy(), None
+
+        canonical_to_csv = {canonical: csv for csv, canonical in column_mapping.items()}
+        enrollment_col = canonical_to_csv.get("Total_Enrollment")
+        crn_col = canonical_to_csv.get("Course_Reference_Number")
+        if not enrollment_col or not crn_col:
+            return courses_df.copy(), None
+
+        col_defs = {cd.canonical_name: cd for cd in schema}
+        enrollment_transformer = col_defs.get("Total_Enrollment").transformer if col_defs.get("Total_Enrollment") else None
+        crn_transformer = col_defs.get("Course_Reference_Number").transformer if col_defs.get("Course_Reference_Number") else None
+
+        enrollment_series = courses_df[enrollment_col]
+        if enrollment_transformer:
+            enrollment_series = enrollment_series.apply(enrollment_transformer)
+
+        # Keep only nonzero enrollments; treat None/NaN as zero for this filter.
+        try:
+            nonzero_mask = enrollment_series.fillna(0).astype(int) != 0
+        except Exception:
+            nonzero_mask = enrollment_series.fillna(0) != 0
+
+        filtered_df = courses_df.loc[nonzero_mask].copy()
+
+        crn_series = filtered_df[crn_col]
+        if crn_transformer:
+            crn_series = crn_series.apply(crn_transformer)
+
+        allowed_crns = {crn for crn in crn_series.tolist() if crn}
+        return filtered_df, allowed_crns
+
+    def _filter_by_allowed_crns(
+        self, enrollments_df: pd.DataFrame, allowed_crns: set[str]
+    ) -> pd.DataFrame:
+        """
+        Filter enrollments to only those whose CRN is in allowed_crns.
+
+        This keeps enrollments consistent with a temporarily filtered course list.
+        """
+        if not allowed_crns:
+            return enrollments_df.copy()
+
+        try:
+            schema, column_mapping = CSVSchemaDetector.detect_schema_version(
+                enrollments_df, "enrollments"
+            )
+        except Exception:
+            return enrollments_df.copy()
+
+        canonical_to_csv = {canonical: csv for csv, canonical in column_mapping.items()}
+        crn_col = canonical_to_csv.get("Course_Reference_Number")
+        if not crn_col:
+            return enrollments_df.copy()
+
+        col_defs = {cd.canonical_name: cd for cd in schema}
+        crn_transformer = col_defs.get("Course_Reference_Number").transformer if col_defs.get("Course_Reference_Number") else None
+
+        crn_series = enrollments_df[crn_col]
+        if crn_transformer:
+            crn_series = crn_series.apply(crn_transformer)
+
+        mask = crn_series.isin(allowed_crns)
+        return enrollments_df.loc[mask].copy()
+
     async def _download_and_parse(self, file_entry: dict) -> tuple[str, pd.DataFrame]:
         """Download one file and parse it."""
         file_type = file_entry["type"]
diff --git a/backend/src/services/schedule/service.py b/backend/src/services/schedule/service.py
index f888f72..b09db1d 100644
--- a/backend/src/services/schedule/service.py
+++ b/backend/src/services/schedule/service.py
@@ -110,6 +110,9 @@ async def generate_schedule(
             # 3. Load course merges (if any) - synchronous call
             merges = self.dataset_service.get_merges(dataset_id, user_id) or {}
 
+            # 3.5 Drop zero-enrollment courses and get updated merges
+            merges = self.dataset_service.drop_zero_enrollment(dataset_id, user_id)
+
             # 4. Build scheduling dataset and run algorithm
             scheduling_dataset = DatasetFactory.from_dataframes_to_scheduling_dataset(
                 courses_df=files["courses"],