diff --git a/src/main/scala/com/fulcrumgenomics/umi/DuplexConsensusCaller.scala b/src/main/scala/com/fulcrumgenomics/umi/DuplexConsensusCaller.scala
index d402cfe40..00434b71c 100644
--- a/src/main/scala/com/fulcrumgenomics/umi/DuplexConsensusCaller.scala
+++ b/src/main/scala/com/fulcrumgenomics/umi/DuplexConsensusCaller.scala
@@ -340,11 +340,16 @@ class DuplexConsensusCaller(override val readNamePrefix: String,
   private[umi] def duplexConsensus(ab: Option[VanillaConsensusRead],
                                    ba: Option[VanillaConsensusRead],
                                    sourceReads: Seq[SourceRead]): Option[DuplexConsensusRead] = {
-    (ab, ba) match {
+    // Calculate the length that we'll create a duplex read as "the shorter of the available reads",
+    // and then filter each read to make sure it has some coverage in that section of the read
+    val len = min(ab.map(_.length).getOrElse(Int.MaxValue), ba.map(_.length).getOrElse(Int.MaxValue))
+    val abX = ab.filter(_.depths.iterator.take(len).exists(_ > 0))
+    val baX = ba.filter(_.depths.iterator.take(len).exists(_ > 0))
+
+    (abX, baX) match {
       case (Some(a), None)    => Some(DuplexConsensusRead(id=a.id, a.bases, a.quals, a.errors, a, None))
       case (None, Some(b))    => Some(DuplexConsensusRead(id=b.id, b.bases, b.quals, b.errors, b, None))
       case (Some(a), Some(b)) =>
-        val len = min(a.length, b.length)
         val id  = a.id
         val bases  = new Array[Byte](len)
         val quals  = new Array[Byte](len)
diff --git a/src/test/scala/com/fulcrumgenomics/umi/DuplexConsensusCallerTest.scala b/src/test/scala/com/fulcrumgenomics/umi/DuplexConsensusCallerTest.scala
index 25047dcbc..e51fad2bd 100644
--- a/src/test/scala/com/fulcrumgenomics/umi/DuplexConsensusCallerTest.scala
+++ b/src/test/scala/com/fulcrumgenomics/umi/DuplexConsensusCallerTest.scala
@@ -28,8 +28,9 @@ import com.fulcrumgenomics.testing.SamBuilder.{Minus, Plus}
 import com.fulcrumgenomics.testing.{SamBuilder, UnitSpec}
 import com.fulcrumgenomics.umi.ConsensusTags.PerRead.{AbRawReadErrorRate, BaRawReadErrorRate, RawReadErrorRate}
 import com.fulcrumgenomics.util.NumericTypes.PhredScore
+import org.scalatest.OptionValues
 
-class DuplexConsensusCallerTest extends UnitSpec {
+class DuplexConsensusCallerTest extends UnitSpec with OptionValues {
   // Function to create a caller without so much writing
   def caller(q: Int = 10, pre: Int = DuplexConsensusCaller.ErrorRatePreUmi, post: Int = DuplexConsensusCaller.ErrorRatePostUmi, minReads: Seq[Int] = Seq(1)) =
     new DuplexConsensusCaller(readNamePrefix="test", minInputBaseQuality=q.toByte, errorRatePreUmi=pre.toByte, errorRatePostUmi=post.toByte,
@@ -468,4 +469,29 @@ class DuplexConsensusCallerTest extends UnitSpec {
       }
     }
   }
+  "DuplexConsensusCaller.duplexConsensus" should "swap AB and BA reads when AB reads have zero depth across the minimum length" in {
+    val cc = caller(minReads=Seq(1, 1, 0))
+    val ab = VanillaConsensusRead(
+      id     = "ab",
+      bases  = "NNN".getBytes,
+      quals  = "III".getBytes,
+      depths = Array(0, 0, 0),
+      errors = Array(0, 0, 0)
+    )
+    val ba   = ab.copy(depths=Array(1, 1, 1))
+
+    // ab has zero depth, so it should be swapped with BA
+    {
+      val read = cc.duplexConsensus(ab=Some(ab), ba=Some(ba), sourceReads=Seq.empty).value
+      read.abConsensus shouldBe ba
+      read.baConsensus.isEmpty shouldBe true
+    }
+
+    // ba has zero depth, so it should be undefined.
+    {
+      val read = cc.duplexConsensus(ab=Some(ba), ba=Some(ab), sourceReads=Seq.empty).value
+      read.abConsensus shouldBe ba
+      read.baConsensus.isEmpty shouldBe true
+    }
+  }
 }