diff --git a/src/imitation/algorithms/adversarial/common.py b/src/imitation/algorithms/adversarial/common.py
index 48129fa67..da578135f 100644
--- a/src/imitation/algorithms/adversarial/common.py
+++ b/src/imitation/algorithms/adversarial/common.py
@@ -176,7 +176,7 @@ def __init__(
                 training. If True, overrides this safety check. WARNING: variable
                 horizon episodes leak information about the reward via termination
                 condition, and can seriously confound evaluation. Read
-                https://imitation.readthedocs.io/en/latest/guide/variable_horizon.html
+                https://imitation.readthedocs.io/en/latest/main-concepts/variable_horizon.html
                 before overriding this.
 
         Raises:
diff --git a/src/imitation/algorithms/base.py b/src/imitation/algorithms/base.py
index fd33c5f40..b1a2b6861 100644
--- a/src/imitation/algorithms/base.py
+++ b/src/imitation/algorithms/base.py
@@ -48,7 +48,7 @@ def __init__(
                 training. If True, overrides this safety check. WARNING: variable
                 horizon episodes leak information about the reward via termination
                 condition, and can seriously confound evaluation. Read
-                https://imitation.readthedocs.io/en/latest/getting-started/variable-horizon.html
+                https://imitation.readthedocs.io/en/latest/main-concepts/variable_horizon.html
                 before overriding this.
         """
         self._logger = custom_logger or imit_logger.configure()
@@ -61,8 +61,8 @@ def __init__(
                 "Additionally, even unbiased algorithms can exploit "
                 "the information leak from the termination condition, "
                 "producing spuriously high performance. See "
-                "https://imitation.readthedocs.io/en/latest/getting-started/"
-                "variable-horizon.html for more information.",
+                "https://imitation.readthedocs.io/en/latest/main-concepts/"
+                "variable_horizon.html for more information.",
             )
         self._horizon = None
 
@@ -100,8 +100,8 @@ def _check_fixed_horizon(self, horizons: Iterable[int]) -> None:
                 f"Episodes of different length detected: {horizons}. "
                 "Variable horizon environments are discouraged -- "
                 "termination conditions leak information about reward. See "
-                "https://imitation.readthedocs.io/en/latest/getting-started/"
-                "variable-horizon.html for more information. "
+                "https://imitation.readthedocs.io/en/latest/main-concepts/"
+                "variable_horizon.html for more information. "
                 "If you are SURE you want to run imitation on a "
                 "variable horizon task, then please pass in the flag: "
                 "`allow_variable_horizon=True`.",
@@ -152,7 +152,7 @@ def __init__(
                 training. If True, overrides this safety check. WARNING: variable
                 horizon episodes leak information about the reward via termination
                 condition, and can seriously confound evaluation. Read
-                https://imitation.readthedocs.io/en/latest/getting-started/variable-horizon.html
+                https://imitation.readthedocs.io/en/latest/main-concepts/variable_horizon.html
                 before overriding this.
         """
         super().__init__(
diff --git a/src/imitation/algorithms/density.py b/src/imitation/algorithms/density.py
index fcc5e5ac9..c59f80c8a 100644
--- a/src/imitation/algorithms/density.py
+++ b/src/imitation/algorithms/density.py
@@ -104,7 +104,7 @@ def __init__(
                 training. If True, overrides this safety check. WARNING: variable
                 horizon episodes leak information about the reward via termination
                 condition, and can seriously confound evaluation. Read
-                https://imitation.readthedocs.io/en/latest/guide/variable_horizon.html
+                https://imitation.readthedocs.io/en/latest/main-concepts/variable_horizon.html
                 before overriding this.
         """
         self.is_stationary = is_stationary
diff --git a/src/imitation/algorithms/preference_comparisons.py b/src/imitation/algorithms/preference_comparisons.py
index 413cd979a..96cc726de 100644
--- a/src/imitation/algorithms/preference_comparisons.py
+++ b/src/imitation/algorithms/preference_comparisons.py
@@ -1550,7 +1550,7 @@ def __init__(
                 training. If True, overrides this safety check. WARNING: variable
                 horizon episodes leak information about the reward via termination
                 condition, and can seriously confound evaluation. Read
-                https://imitation.readthedocs.io/en/latest/guide/variable_horizon.html
+                https://imitation.readthedocs.io/en/latest/main-concepts/variable_horizon.html
                 before overriding this.
             rng: random number generator to use for initializing subcomponents such as
                 fragmenter.
diff --git a/src/imitation/scripts/train_preference_comparisons.py b/src/imitation/scripts/train_preference_comparisons.py
index 79ee4c136..e9703f78b 100644
--- a/src/imitation/scripts/train_preference_comparisons.py
+++ b/src/imitation/scripts/train_preference_comparisons.py
@@ -133,7 +133,7 @@ def train_preference_comparisons(
             training. If True, overrides this safety check. WARNING: variable
             horizon episodes leak information about the reward via termination
             condition, and can seriously confound evaluation. Read
-            https://imitation.readthedocs.io/en/latest/guide/variable_horizon.html
+            https://imitation.readthedocs.io/en/latest/main-concepts/variable_horizon.html
             before overriding this.
         checkpoint_interval: Save the reward model and policy models (if
             trajectory_generator contains a policy) every `checkpoint_interval`