diff --git a/src/staar/output.rs b/src/staar/output.rs
index 11d73e9..a333d6e 100644
--- a/src/staar/output.rs
+++ b/src/staar/output.rs
@@ -103,9 +103,11 @@ pub struct IndividualRow {
 pub fn write_individual_results(
     rows: &[IndividualRow],
     output_dir: &Path,
+    is_conditional: bool,
     out: &dyn Output,
 ) -> Result<(), CohortError> {
-    let out_path = output_dir.join("individual.parquet");
+    let stem = if is_conditional { "individual_cond" } else { "individual" };
+    let out_path = output_dir.join(format!("{stem}.parquet"));
     let n = rows.len();
 
     // R orders by POS ascending before returning (Individual_Analysis.R:450).
@@ -381,8 +383,10 @@ pub fn write_results(
     trait_type: TraitType,
     n: usize,
     n_rare: i64,
+    is_conditional: bool,
     out: &dyn Output,
 ) -> Result<(), CohortError> {
+    let cond_suffix = if is_conditional { "_cond" } else { "" };
     out.status("Writing results...");
     let mut significant_genes: Vec<serde_json::Value> = Vec::new();
 
@@ -397,7 +401,8 @@ pub fn write_results(
         if results.is_empty() {
             continue;
         }
-        let out_path = output_dir.join(format!("{}.parquet", mask_type.file_stem()));
+        let out_path =
+            output_dir.join(format!("{}{cond_suffix}.parquet", mask_type.file_stem()));
 
         let nan_pvals = results.iter().filter(|r| r.staar.staar_o.is_nan()).count();
         if nan_pvals > 0 {
@@ -453,6 +458,7 @@ pub fn write_results(
         "cohort_staar_version": 1,
         "traits": trait_names, "trait_type": format!("{:?}", trait_type),
         "n_samples": n, "n_rare_variants": n_rare, "maf_cutoff": maf_cutoff,
+        "conditional": is_conditional,
         "significant_genes": significant_genes,
     });
     // Merge null-model summary fields (sigma2 for single-trait, or the
diff --git a/src/staar/pipeline.rs b/src/staar/pipeline.rs
index fbedcc0..6d33ed8 100644
--- a/src/staar/pipeline.rs
+++ b/src/staar/pipeline.rs
@@ -825,6 +825,7 @@ impl<'a> StaarPipeline<'a> {
             TraitType::Continuous,
             n,
             n_rare,
+            self.config.known_loci.is_some(),
             self.out,
         )
     }
@@ -986,8 +987,15 @@ impl<'a> StaarPipeline<'a> {
             ))
         })?;
 
+        let is_conditional = self.config.known_loci.is_some();
+
         if self.config.individual && !scoring.individual.is_empty() {
-            write_individual_results(&scoring.individual, &results_dir, self.out)?;
+            write_individual_results(
+                &scoring.individual,
+                &results_dir,
+                is_conditional,
+                self.out,
+            )?;
         }
 
         write_results(
@@ -1001,6 +1009,7 @@ impl<'a> StaarPipeline<'a> {
             trait_type,
             n,
             n_rare,
+            is_conditional,
             self.out,
         )
     }
@@ -1019,6 +1028,11 @@ impl<'a> StaarPipeline<'a> {
                 "--spa with --kinship: SPA is applied at the score-test layer; the kinship-aware path provides exact variance via the GLMM projection.",
             );
         }
+        if self.config.known_loci.is_some() && mode == ScoringMode::Spa {
+            self.out.warn(
+                "--spa with --known-loci: the known-loci adjustment is absorbed into the fitted null, but the SPA saddlepoint reads raw G; gene-mask p-values from this run are SPA-adjusted but not conditional. A separate --cond-spa path is tracked in STAARpipeline as Gene_Centric_*_cond_spa.",
+            );
+        }
     }
 
     fn cohort(&self) -> CohortHandle<'_> {